ArticlePDF AvailableLiterature Review

Abstract and Figures

The present model outlines the mechanisms underlying habitual control of responding and the ways in which habits interface with goals. Habits emerge from the gradual learning of associations between responses and the features of performance contexts that have historically covaried with them (e.g., physical settings, preceding actions). Once a habit is formed, perception of contexts triggers the associated response without a mediating goal. Nonetheless, habits interface with goals. Constraining this interface, habit associations accrue slowly and do not shift appreciably with current goal states or infrequent counterhabitual responses. Given these constraints, goals can (a) direct habits by motivating repetition that leads to habit formation and by promoting exposure to cues that trigger habits, (b) be inferred from habits, and (c) interact with habits in ways that preserve the learned habit associations. Finally, the authors outline the implications of the model for habit change, especially for the self-regulation of habit cuing.
Content may be subject to copyright.
A New Look at Habits and the Habit–Goal Interface
Wendy Wood and David T. Neal
Duke University
The present model outlines the mechanisms underlying habitual control of responding and the ways in
which habits interface with goals. Habits emerge from the gradual learning of associations between
responses and the features of performance contexts that have historically covaried with them (e.g.,
physical settings, preceding actions). Once a habit is formed, perception of contexts triggers the
associated response without a mediating goal. Nonetheless, habits interface with goals. Constraining this
interface, habit associations accrue slowly and do not shift appreciably with current goal states or
infrequent counterhabitual responses. Given these constraints, goals can (a) direct habits by motivating
repetition that leads to habit formation and by promoting exposure to cues that trigger habits, (b) be
inferred from habits, and (c) interact with habits in ways that preserve the learned habit associations.
Finally, the authors outline the implications of the model for habit change, especially for the self-
regulation of habit cuing.
Keywords: habit, goal, automaticity, behavior change, self-regulation
Most of the time what we do is what we do most of the time.
Sometimes we do something new (Townsend & Bever, 2001, p. 2).
From the humdrum to the consequential, daily actions tend to be
patterned into sequences that are repeated at particular times in
customary places. If Townsend and Bever (2001) are correct, the
majority of day-to-day living is characterized by repetition in this
Empirical estimates of repetition in daily life come from signal-
contingent experience-sampling diary investigations. Participants
in these studies recorded once per hour for several days what they
were doing, thinking, and feeling (Quinn & Wood, 2005; Wood,
Quinn, & Kashy, 2002). In college student as well as community
samples, about 45% of the behaviors participants listed in their
diaries tended to be repeated in the same physical location almost
every day. Substantial amounts of repetition in stable contexts also
have been documented with other naturalistic paradigms. In Barker
and Schoggen’s (1978) ecological analysis, observers from the
Midwest Psychological Field Station obtained finely detailed re-
cordings of children’s everyday activities in a small town. The
researchers found a high degree of repetition in daily activities, and
consistent with the diary studies, this repetition was linked to
specific environments. Accordingly, Barker (1968) proposed that
the most proximal predictor of responding is the behavior setting,
defined as “standing patterns of behavior-and-milieu” (p. 19).
Why do people repeat actions in contexts in this way? In the
heyday of behaviorism, psychologists invoked associative learning
mechanisms and stimulus–response (S-R) habits to explain re-
peated responding cued by recurring stimuli. More recently, social
and personality psychologists have attributed consistency in re-
sponding to people’s goals, intentions, and other dispositions (e.g.,
attitudes, personality) that lead them to value, and hence to pursue
repeatedly, particular outcomes in particular contexts. In this arti-
cle, we outline a synthetic theory that integrates habit responding
with recognition of the essentially goal-directed nature of much
human action. As we show below, habits are neither the simple
S-R links advanced by some behaviorists nor the automatic ex-
pression of people’s goals. In our model, habits are subserved by
a form of automaticity that involves the direct association between
a context and a response but that interfaces with goals during
learning and performance.
New Model of Habits in Brief
Habits are learned dispositions to repeat past responses. They
are triggered by features of the context that have covaried fre-
quently with past performance, including performance locations,
preceding actions in a sequence, and particular people. Contexts
activate habitual responses directly, without the mediation of goal
states. We decompose this definition into three principles that play
out in the acquisition of habits and in their performance once
The first principle in our model centers on the power of contexts
to trigger habitual responding. That is, the automaticity underlying
habits builds on patterns of covariation between features of per-
formance contexts and responses—patterns that arise intentionally
or unintentionally in the course of daily life. People form habits as
they encode these context–response patterns in procedural mem-
ory. Once formed, the habitual response comes to be primed or
Wendy Wood and David T. Neal, Department of Psychology and
Neuroscience, Duke University.
We thank Jeffrey Quinn for his important contribution to early stages of
this work. In addition, the manuscript was improved from the thoughtful
commentary of Henk Aarts, Dolores Albarracı´n, John Bargh, Marilynn
Brewer, Tanya Chartrand, Joel B. Cohen, Mark Conner, Anthony Dickin-
son, Alice H. Eagly, Scott Huettel, Bas Verplanken, and the students in the
Duke Interdisciplinary Initiative in Social Psychology. This work was
supported by the Social Science Research Institute at Duke University.
Correspondence concerning this article should be addressed to Wendy
Wood, Department of Psychology and Neuroscience, Duke University,
Box 90085, Durham, NC 27708. E-mail:
Psychological Review Copyright 2007 by the American Psychological Association
2007, Vol. 114, No. 4, 843– 863 0033-295X/07/$12.00 DOI: 10.1037/0033-295X.114.4.843
triggered by the perception of cues in the performance context.
This process could be initiated by entering the physical setting in
which the habit typically is performed, completing the response
that typically precedes it, or encountering a person who typically
is present. Thus, the first principle stipulates an outsourcing of
behavioral control to context cues that were, in the past, contigu-
ous with performance.
We propose that the context cuing of habits arises in two
possible forms. In the direct form, habit responding is activated by
cold cognitive associations between context cues and responses,
and in the motivated form, it is activated by the diffuse motivation
that is tagged onto performance contexts when people repeatedly
experience rewards for responding in those contexts.
The second principle concerns the absence of goal mediation in
the context–response associations that make up habits. Habits
typically are the residue of past goal pursuit; they arise when
people repeatedly use a particular behavioral means in particular
contexts to pursue their goals. However, once acquired, habits are
performed without mediation of a goal to achieve a particular
outcome or a goal to respond (i.e., a behavioral intention). For
example, purchase of a particular newspaper each morning with
coffee initially is guided by the recruitment of a mental represen-
tation of a goal (e.g., acquiring information, wasting time). How-
ever, recruitment of this goal becomes progressively less necessary
as newspaper purchase is repeated and becomes integrated with the
morning coffee-purchase routine so that it can be triggered by
relevant context cues (e.g., sight of the barista, act of ordering
coffee). The second principle thus differentiates habits from more
flexible, goal-dependent forms of automaticity, in which goals
continue to mediate responses albeit in a manner that is automatic,
implicit, and/or nonconscious. As we explain below, the distinc-
tion between these two forms of automaticity is evident at the
behavioral level. Habitual responses are triggered rigidly by asso-
ciated cues in the performance context, whereas responses guided
by implicit goals are performed more variably and often depend on
the activation of supporting explicit goals.
The first two habit principles largely are compatible with folk
concepts of habit performance. “I can’t help it, it’s just a habit,” is
an excuse that people might offer for such cued behaviors as bad
habits (e.g., chronic overeating) and action slips (e.g., accidentally
driving to work when intending to go to the store). By offering
such accounts, people perhaps are acknowledging that their re-
sponses are cued by performance contexts independently of what
they intended to accomplish.
On the basis of the first two principles, habit performance might
seem an obligatory, reflexlike response to associated context cues.
However, the third principle of our model delineates habits’ inter-
face with goals and related dispositions. Habits arise from context–
response learning that is acquired slowly with experience. As a
result, habit dispositions do not alter in response to people’s
current goals or occasional counterhabitual responses. Thus, habits
possess conservative features that constrain their relation with
goals. Within these constraints, goals and habits can direct each
other. As we explain below, people’s goals can guide the forma-
tion of habit associations, and people can rely on habits to make
inferences about their goals. Furthermore, when habits and goals
are both present to guide action, they interact in their effects such
that under some circumstances people respond habitually and
under others they exert regulatory control to inhibit the cued
response and perhaps perform a more desired one. Understanding
of the habit– goal interface thus echoes dual-mode models that
specify how automatic, associative processes interact with more
controlled, rule-based processes (see Chaiken & Trope, 1999;
Smith & DeCoster, 2000).
These three principles provide the foundation for a host of new
research questions. As we speculate at the end of this article, the
habit– goal interface allows for established habits to be co-opted in
the service of goals different from those that directed habit forma-
tion. Also, some simple forms of habit regulation might proceed
without the use of goals as comparison standards. The model also
offers new insights into established literatures, especially how to
tailor behavior change interventions to maximize their impact on
Habits in Historical Perspective
Our model of habits and their interface with goals synthesizes
diverse theoretical traditions that influenced theories of action
control across the last century. The habit construct has strong
historical ties to behaviorism, especially to Watson’s (1913) and
Skinner’s (1938) radical behaviorism that famously eschewed cog-
nitive and motivational mediators of behavior. These forms of
behaviorism built closely on Thorndike’s (1898) notion of learning
as the formation of a direct bond between some physical event or
sensory input and a muscle response, so that the external stimula-
tion reflexively comes to cause the response. The decline of this
perspective in psychology is often traced to Chomsky’s (1959) and
Mowrer’s (1960) famous critiques that highlighted the inadequacy
of radical behaviorism’s reduction of complex human behavior,
especially speech production and language, to a linear series of
single S-R units.
As the limitations of some behaviorist models of human func-
tioning became apparent, a new era of cognitive science developed
that rejected central assumptions of those models. Instead of lo-
cating the cause for behavior in the environment, the new perspec-
tives situated causality in internal mental processes, specifically, in
a hypothesized central executive controller (see Neisser, 1967).
Cognitive science researchers also raised questions about the ad-
equacy of associationism, central to behaviorist logic, in which
sets of words, items, or mental representations can become asso-
ciated through bottom-up processes in which perception of one
element produces, generates, or arouses the other. With the view
that associationism is “reductionist and mechanical and not in
keeping with apparent complexities of human memory” (Mandler,
2002, p. 334), cognitive science models shifted emphasis to focus
on the purposive, top-down organization of perceptions and con-
In the past decade, a broad social– cognitive– behaviorist syn-
thesis has developed that incorporates key elements of the behav-
iorists’ toolbox within a framework of the inherently goal-directed
nature of human action. As Bargh and Ferguson (2000) noted,
cognitive models of executive control have been extended to
include a favored behaviorist tool, the causal role of the environ-
ment. For example, in social cognitive models of automaticity,
goal-dependent responding can be triggered by environmental
stimuli. Another useful tool for behaviorism, learning through
association, ultimately never fell out of favor among cognitive
scientists. The formation of simple associations between individual
concepts provides a foundation for connectionism and other cog-
nitive theories of the development of large, complex systems of
meanings (see Bower, 2000). Within social psychology, the logic
of associationism forms the basis for many of the low effort,
heuristic processes in persuasion, stereotyping, and person percep-
tion (see Smith & DeCoster, 2000). The present model further
develops the emerging synthesis. It retains key features of habit
mechanisms postulated by early S-R theorists, while drawing
habits into dialogue with goal systems.
Habit Responses Are Cued by Contexts Without
Mediation by Goals
In our model, habits are repeated responses that come to be cued
by recurring features of contexts (Principle 1) without mediation
by a cognitive representation of a goal (Principle 2). Although we
treat these principles as distinct for analytical purposes, we regard
them ultimately as dual facets of a unified habit disposition. Each
principle is necessary, and by itself, neither is sufficient to define
habits. In this way, our definition aligns with classic treatments of
habit dispositions. Most notably, William James (1890) proposed
that habits are triggered spontaneously by sensory cues and pre-
ceding actions, and that this cuing proceeds without recourse to
goal-related constructs of volition and will.
In the next sections, we evaluate the empirical evidence for each
principle separately as this evidence emerges across the relevant
research literatures. Principle 1 situates habit cuing within the
broader idea that responses can come to be triggered by context
cues that have reliably accompanied prior performance. The em-
pirical evidence for this facet of habitual responding comes pre-
dominantly from cognitive and neural models of the learning
mechanisms that underlie repeated responding in contiguity with
particular context cues. Principle 2 differentiates habit cuing from
other goal-dependent forms of automaticity. The empirical evi-
dence for this facet of habitual responding comes primarily from
behavior prediction research, computational models of routine
action, neuroimaging studies of automaticity, and animal learning
Principle 1: Habits are cued by context.
Context cues refer broadly to the many elements of the perfor-
mance environment that potentially can recur as actions are re-
peated, including physical locations, other people, and preceding
actions in a sequence. The first principle of our model reflects the
generally accepted ideas that humans, like other animals, are adept
at detecting these patterns of covariation and encoding them in
mental representations that chunk contexts and responses. Con-
texts can then activate directly—that is, automatically and without
recruitment of a mediating goal—performance of the response.
This outsourcing of behavioral control to context cues captures the
essence of Principle 1.
The power of contexts to cue habit responding is evident in both
laboratory and naturalistic paradigms (see Neal & Wood, in press).
In laboratory settings, the historical covariation of context cues and
responses is measured or manipulated in order to demonstrate the
facilitating effects of contexts on the speed or accuracy of respond-
ing. Studies in this vein show that people encode and exploit,
sometimes without conscious awareness, context–response co-
variation that is based on abstract visual cues (e.g. LaBar & Phelps,
2005; Lewicki, Hill, & Bizot, 1988), prior responses in learned
sequences (see Graybiel, 1998), and everyday objects associated
with particular physical motions (e.g., power gripping of hammers,
precision gripping of keys; Tucker & Ellis, 2004).
As we noted in the introduction to this article, naturalistic
paradigms reveal substantial covariation between contexts and
responses that can form the basis for outsourcing behavioral con-
trol. Naturalistic data also suggest that this context–response co-
variation confers causal power onto contexts in activating re-
sponses. In this regard, Wood, Tam, and Guerrero Witt (2005)
studied everyday habits (e.g., reading the newspaper) of college
students transferring to a new university. Suggesting the causal
power of contexts, students’ habits were disrupted when the trans-
fer altered specific features of the performance context for that
behavior, and furthermore, students’ goals to respond were not
able to account for this disruption.
In sum, a variety of responses can be primed by a range of cues
that have in the past reliably covaried with the response. Despite
the extensive evidence that such cuing occurs, research has yet to
identify the exact psychological mechanisms through which con-
texts activate associated overt responses. We consider below two
possible forms of this cuing, which we call direct cuing and
motivated cuing (see Neal, Wood, & Quinn, 2006). As we explain
below, both provide promising, although as of yet somewhat
unexplored, accounts of how people’s perception of context cues
can produce habit performance. Understanding of habit cuing
mechanisms is not yet sufficient to prefer one account over the
other, and it may be that both function to some extent to promote
habit performance.
Direct Cuing
When directly cued, habits are represented in memory as direct
context–response associations that develop from repeated coacti-
vation of the context and response. That is, when the mental
representation of a response (e.g., buckling seatbelt) is consistently
activated in conjunction with representation of a context (e.g.,
getting into a car), associative links gradually form between the
two (e.g., buckling seatbelt entering a car).
The essential mechanism behind direct cuing involves the cog-
nitive neural changes that result from repeated coactivation of
responses and contexts (see Hebb, 1949). With repetition, incre-
mental changes occur in relevant processors or neural assemblies
in procedural memory, essentially tuning the processing elements
in ways that facilitate the repeated aspects of responding to recur-
ring features of performance contexts. This gradual development
over repeated experience provides a selection mechanism for habit
learning because only those patterns that are consistently and
frequently repeated will be encoded in procedural memory in the
form of habit associations (see Gupta & Cohen, 2002; McClelland,
McNaughton, & O’Reilly, 1995).
Simple coactivation plausibly explains how the perception of
context cues activates a mental representation of a historically
associated response. However, it is less clear how simple height-
ened cognitive accessibility drives overt habit responses. In one
account, the activation of responding emerges via an ideomotor
mechanism, which stipulates that the mere thought of a behavior
tends to lead to performance of it (James, 1890). Thus, a context
cue may directly activate an associated response via simple asso-
ciative learning, and this activated response may then be enacted
via an ideomotor mechanism. In a series of compelling demon-
strations of ideomotor effects, Bargh, Dijksterhuis, and colleagues
have shown that participants primed with the elderly stereotype
walked slowly (Bargh, Chen, & Burrows, 1996), generated slow
response latencies (Dijksterhuis, Spears, & Lepinasse, 2001), and
displayed poor memory (Dijksterhuis, Bargh, & Miedema, 2000).
However, in this research, activation of stereotypes and concepts
by features of contexts influenced the expression but not neces-
sarily the initiation of responding. As Bargh et al. (1996) noted,
activation of the elderly stereotype decreased walking speed of
participants who had already chosen to walk down the hall but did
not lead them to initiate a new stereotype-linked response (e.g.,
buying condos in Florida).
The effects of concept activation can be compared with
mimicry-based ideomotor movement effects that, in contrast, have
been shown to initiate overt responding. For example, observation
of other people’s movements appears to produce unintentional,
nonconscious mimicry as a result of a common neural substrate
that supports the perception and performance of action (i.e., mirror
neurons, Rizzolatti, Fadiga, Gallese, & Fogassi, 1996; see also
Chartrand & Bargh, 1999; Prinz, 1990). However, mimicry effects
do not provide a sufficient account for habit cuing, given that
habits can be triggered not only by movement but also by repre-
sentations of such varied cues as locations, simple presence of
others, and preceding responses that occurred contiguous with a
response in the past.
Thus, outside of ideomotor effects that arise from a common
neural substrate for perceiving and acting, the direct cuing effects
that have been identified to this point largely involve action
representations influencing the form, or manner, of consciously
intended action. It remains to be demonstrated whether the simple
coactivation in direct cuing (e.g., representations of popcorn
movie theater) provides a sufficient impetus to initiate an overt
habit response (e.g., actually purchasing the popcorn) as opposed
to modifying an already initiated stream of action.
Motivated Cuing
Habit associations also may arise through a process in which the
reward value of response outcomes (e.g., positive affect from
eating popcorn) is conditioned onto context cues (e.g., movie
theater) that have historically accompanied the receipt of those
rewards. We refer to this process as motivated cuing because
context cues in this case carry hot motivational influence insofar as
they signal opportunities to perform rewarded responses (see Neal
et al., 2006). This analysis builds on instrumental learning theories
in which habits develop as organisms learn context–response as-
sociations in order to obtain rewarding events (see Dickinson &
Balleine, 2002). In brief, the logic of incentive conditioning sug-
gests that when cues in the performance context are contiguous
with a rewarded response, the reward value becomes conditioned
onto the cues. Given sufficient repetition, the cues themselves then
carry power to motivate the response because they signal an
opportunity to acquire the associated reward. In this account then,
contexts drive habit performance because past reward conditioning
not only establishes cognitive context–response associations but
also imbues the context with the motivational impetus for respond-
ing. Thus, it is possible that motivational cuing works to augment
and enhance, rather than replace, context–response learning based
on direct cuing.
Models of how incentive conditioning can promote habit per-
formance are developing in research on the neurotransmitter sys-
tems that scaffold response to reward. To illustrate, we focus on
the neurotransmitter dopamine, although the full story of moti-
vated cuing requires more than our necessarily brief account of
dopaminergic function (see Dayan & Balleine, 2002). Dopamine
acts in the nucleus accumbens, dorsal and ventral striatum, amyg-
dala, frontal cortex, and perhaps other sites to promote learning of
rewards to guide future behavior. Dopamine responses reflect a
sort of prediction error sensitive to differences between expected
and obtained rewards and to future expectations of reward (Mon-
tague, Hyman, & Cohen, 2004; Schultz, 2006). Thus, midbrain
neurons in dopamine-rich areas emit a positive signal of brief,
spiked activation to unexpectedly large rewards, limited or no
activation to expected rewards, and a negative signal of decreased
activity to unexpectedly small or absent rewards.
Evidence of dopamine’s role in habit learning and performance
comes largely from nonhuman research. However, the extension to
humans is plausible given that habit performance involves phylo-
genetically primitive learning mechanisms that likely are shared
across mammalian species. In support, neuroscientists have noted
reassuringly equivalent forms of habit learning in humans and
experimental animals that are mediated by common brain struc-
tures involving the basal ganglia (e.g., Packard & Knowlton,
2002). Moreover, human imaging studies have demonstrated do-
paminergic responses to abstract rewards of money as well as to
appetitive rewards of food (O’Doherty, 2004). At least with re-
spect to appetitive rewards, this activation is similar to that found
with nonhuman samples (Montague et al., 2004).
Dopamine response systems appear to make multiple contribu-
tions to habit formation and performance, including in the initial
stamping-in of context–response associations in memory and in
maintaining the incentive value of established context–response
mappings. These dual roles are illustrated in the often-cited study
by Mirenowicz and Schultz (1996) in which monkeys initially
learned a task in which an environmental stimulus (e.g., a light)
predicted a reward (e.g., a drop of juice) when they gave a
response (e.g., pressed a lever). At the beginning of learning,
activity in dopamine-rich areas of the monkey’s brain occurred just
after receipt of the reward. Dopamine activation following a be-
havior that is being reinforced works retroactively to stamp in the
still-active memory traces of the stimulus and the response. In this
way, dopamine augments learning of context–response associa-
tions (see Wise, 2004). After several days of training, the animals
had learned the task, and they reached for the lever as soon as the
light was illuminated. Note that this repetition also shifted dopa-
mine responding so that its effects were apparent proactively. That
is, the dopamine response was no longer elicited by the reward
itself but instead was activated to the earliest cue to the reward, the
initial presentation of the light. Thus, with repetition, dopamine
responses appeared to transfer from rewards to reward-predictive
context cues.
Although the mechanisms of the transfer of dopamine respond-
ing onto contexts have yet to be demonstrated with humans,
neuroimaging studies have revealed a parallel phenomenon in
which, as people practice probabilistic instrumental tasks, brain
activation in dopaminergic areas increases in response to cues that
signal opportunities to perform rewarded responses (see Knutson
& Cooper, 2005). In addition, in research on dopamine-altering
drugs like cocaine, incidental cues in the environments accompa-
nying people’s past drug use apparently can reinstate the drug
craving (see Kalivas & McFarland, 2003). In these ways, perfor-
mance environments in which goals are reached or rewards are
received may acquire the capacity to motivate historically associ-
ated responses.
How does this motivational account of context cuing effects
mesh with our claim that the context cuing underlying habits is not
goal mediated? The motivational substrate for habit responding
has unique properties that stem from it being a cumulative residue
of consistently rewarded responding. That is, the motivation trans-
ferred onto predictive environmental cues is an accumulated value
that is encoded as a part of the context in the learning of context–
response associations and does not vary flexibly with changes in
the current outcomes of responding (Daw, Niv, & Dayan, 2005).
As a result, context cues provide only a relatively diffuse motiva-
tion for habit performance. As Daw et al. (2005) proposed, when
habits form, the performance contexts of reward become associ-
ated with a cached value representing the summary of their long-
run future value. This cached value develops through procedural
learning and is supposedly independent of any specific outcome
information. Thus, the motivational impetus that comes from in-
centive conditioning can be differentiated from goals, at least to
the extent that goals represent particular desired outcomes.
It is worth noting that incentive conditioning onto contexts is not
the only reward mechanism that could underlie habit performance.
Custers and Aarts (2005, 2007) proposed that with conditioning of
positive affect onto cognitive representations of responses, the
response representations acquire motivating properties. They spec-
ulated further that the mechanisms of such conditioning could be
located in dopaminergic responding, as the experience of associ-
ations between behaviors and positive feelings excites brain struc-
tures that encode the behaviors’ general desiredness. Subsequent
activation of such motivated response representations then insti-
gates wanting to perform the response. Although Custers and Aarts
interpreted positive affect as a property of the cognitive represen-
tation of goals, the evidence of the motivating properties of affec-
tive conditioning is equally amenable to a broader interpretation
that does not involve specific goal representations. Within our
habit model, the findings suggest that contexts can activate re-
sponse representations on the basis of learning of contiguity, and
response representations that acquire a broadly positive valence
(similar to a cached value) as a result of affective conditioning can
drive overt performance.
To summarize motivated cuing, models of the neurotransmitter
processes that underlie instrumental learning provide a mechanism
by which the context cues reliably associated with response out-
comes can come to motivate habit performance. Specifically,
models of dopamine function can explain how the diffuse moti-
vating properties of rewards are gradually transferred with habit
formation onto the contexts of instrumental performance. These
contexts then can energize the associated response. Additionally, it
is possible that motivation is conferred onto the response repre-
sentation, so that activating the representation is sufficient to drive
overt performance (see Custers & Aarts, 2005, 2007).
To summarize context cuing in general, the direct and motivated
forms both provide potential accounts of the psychological sub-
strate by which context cues trigger overt habit performance. In
either form, repeated responses can be activated in memory by
associated contexts, and the activated response representations can
drive performance without requiring the mediating involvement of
a goal. As we have characterized them, direct cuing represents a
cold, nonmotivated process, whereas motivated cuing emerges
from the value of the rewarding experiences associated in the past
with contexts and responses. These two forms of habit cuing are
grounded in separate research literatures in psychology, with direct
cuing arising within social cognition and motivated cuing within
neural models of reward learning in animals. Despite their differ-
ences, these forms of cuing both have the potential to provide a
coherent account of the psychological processes that undergird
habit development and performance. Our guess is that future
theorizing on habit cuing will build further on, and perhaps inte-
grate, both mechanisms.
In presenting Principle 1, we described how perception of con-
texts can activate habit responses in memory and promote their
overt performance. In the next section of the article, we explain the
basis for our second principle, that goals do not mediate habit
performance. In making this assertion, we draw on the widely
accepted idea that goals can be represented in memory just as can
other types of information (e.g., Kruglanski et al., 2002).
Principle 2: Habit context–response associations are not me-
diated by goals.
According to our second principle, habit associations are not
mediated by representations of goals. This claim builds on decom-
positional models of automaticity that allow automatic processes
to exhibit multiple separable features that can be present in various
combinations (Bargh, 1994; Moors & De Houwer, 2006). In the
language of these approaches, habits can be categorized as a
goal-independent form of automaticity, given that habit perfor-
mance “does not depend on a goal for its occurrence” (Moors &
De Houwer, 2006, p. 305). However, given that habits typically
originate in goal pursuit, habit performance often inadvertently
promotes goal-consistent outcomes. That is, although they are not
goal mediated, habits may blindly carry out the work of the goal
that initially prompted people to respond repeatedly and thus to
develop the habit. Thus, habits may be goal directed in this
restrictive sense, even though they are not goal dependent.
In contrast to our model, habits sometimes have been defined as
a form of goal-dependent automaticity (e.g., Aarts & Dijksterhuis,
2000; Verplanken & Aarts, 1999). In this alternative view, habits
are represented mentally as goal–action links that emerge when
context cues activate a goal and thereby an associated action to
achieve that goal. However, as we explain, the features of goal-
dependent automaticity do not correspond with the features of
habit performance. In particular, automatic goal pursuit is charac-
terized by variability in response rather than repetition of any
particular behavioral means.
Recognizing the flexibility inherent in automatic goal pursuit,
Bargh and Barndollar (1996) argued that the environmental acti-
vation of goals yields “not a static behavioral response, but an
automated strategy for dealing with the environment” (p. 461,
italics in original). Building on this idea that goals convey a
malleable, dynamic orientation, a principle of goal systems theory
is substitutability in the means of goal pursuit (see Kruglanski et
al., 2002). That is, to the extent that goals possess the property of
equifinality (i.e., goals can be met through multiple means), then it
is plausible that means of comparable expected value can be
substituted for each other. Even strongly desired goals that stably
characterize people’s motives do not necessarily yield stability in
the particular means or responses involved in goal pursuit. Thus,
activating a goal to be healthy might prompt people sometimes to
forgo dessert and other times to take a walk.
Research on automatic goal pursuit provides substantial evi-
dence of variability in responses promoted by the activation of
nonconscious goals. This variability emerges in part because re-
sponses to implicit goals are flexibly moderated by the explicit
goals that people consciously are pursuing. For example, an im-
plicitly activated goal, such as the desire to help others, that is
inconsistent with a conscious goal, such as to be on time, appears
to have little effect on responding (Macrae & Johnston, 1998).
Variability in response also is characteristic of implementation
intentions, a form of automatic goals in which people plan to
perform a particular response upon encountering a particular cue.
For example, Sheeran, Webb, and Gollwitzer (2005, Study 1)
found that participants who had formed implementation intentions
to study at particular places and times acted accordingly only if
they explicitly endorsed the goal of studying. In a second study,
participants were found to act according to their automated inten-
tions to respond quickly at a task only if a broader achievement-
related goal had been primed outside of awareness (Sheeran,
Webb, & Gollwitzer, 2005). Furthermore, in line with the idea that
implicit goal effects depend on people’s explicit goals, the
achievement goal prime in this study failed to influence perfor-
mance speed by itself (i.e., when unaccompanied by an implemen-
tation intention to respond quickly), presumably because all par-
ticipants had a conscious task goal of being accurate, and this
overrode the impact of priming. In explaining this set of findings,
Sheeran, Webb, and Gollwitzer (2005) noted that “the term ‘stra-
tegic’ captures an important feature of the automaticity in imple-
mentation intentions that is different to the automaticity associated
with habits” (p. 96). Thus, automated intentions appear to produce
the associated behavior primarily when supported by explicit
goals. In general, although cases of conflict between explicit and
implicit goals are not always resolved in favor of explicit ones
(e.g., sometimes implicit goals impair explicit goal pursuit; Shah &
Kruglanski, 2002), the dependence between implicit and explicit
goals promotes variability in associated responses.
In sharp contrast to the evidence that automatic goal effects are
variable and depend on explicit goals, habit performance does not
appear to depend on the availability or accessibility of a supporting
goal in memory. Evidence that habits can be performed without an
available supporting explicit goal comes from naturalistic studies
predicting the frequency with which people perform everyday
behaviors such as watching TV, purchasing fast food, driving a
car, and recycling (see review in Ouellette & Wood, 1998; see also
Ji & Wood, in press; Verplanken, Aarts, van Knippenberg, &
Moonen, 1998). In a typical study, a regression model is con-
structed to predict future performance from the favorability of
people’s behavioral goals and the strength of their existing habits
(as reflected in frequency of past performance, see also Verplan-
ken & Orbell, 2003). The standard result is that habit strength is an
independent predictor of the extent to which people repeat activ-
ities. In a detailed exploration of this pattern, Ouellette and Wood
(1998, Study 2) found that habit strength continued to predict
future responses even in study designs that controlled for people’s
(a) explicit behavioral goals, (b) perceptions of efficacy and con-
trol over performance, (c) self-concept as someone who performs
the act or not, and (d) attitude accessibility, as assessed by reaction
times to give attitude judgments. Thus, habit responses continue to
be performed even in the absence of an available supporting
explicit goal or other disposition (e.g., attitude accessibility).
Additional evidence for the idea that habits are not a form of
automatic goal pursuit comes from Neal, Pascoe, and Wood’s
(2007) research that manipulated the accessibility of performance
goals. In this research, participants formed habits in a probabilistic
cue–response task involving the prediction of weather (rain versus
shine) based on geometric shapes on tarot cards (see Knowlton,
Mangels, & Squire, 1996). In the habit formation condition, par-
ticipants learned by first guessing the outcome on each trial and
then being presented with the correct outcome. To encourage
reliance on habit-based procedural memory, participants in this
condition also counted auditory tones. In the control condition,
participants learned by simply observing the tarot cards and the
weather outcome presented simultaneously, thus formulating de-
clarative rules (e.g., circles indicate rain). Prior fMRI studies and
studies with clinical populations have established that the feedback
condition does in fact engage habit-based procedural memory,
whereas the observational condition predominantly engages de-
clarative memory (Knowlton et al., 1996; Poldrack et al., 2001).
To evaluate whether habit performance depends on accessibility
of a supporting goal, Neal et al.’s participants were primed after
the initial learning phase with achievement goals or they were not
primed, and then all participants were tested on their weather-
prediction ability. Those in the control condition, in which learning
involved declarative rules, performed significantly better on the
test when the implicit achievement goal was made accessible (see
similar pattern in Bargh, Gollwitzer, Lee-Chai, Barndollar, &
Tro¨tschel, 2001). These participants apparently applied the task
rules more accurately when they had an enhanced goal to perform
well. However, those in the habit-based procedural learning con-
dition performed significantly worse when the achievement goal
was primed, a phenomenon reminiscent of choking effects on
skilled performance (Baumeister, 1984). These results were repli-
cated in a second study in which the achievement goal was
activated explicitly through performance-contingent payment. The
overall pattern of Neal et al.’s findings suggest that habits are a
form of goal-independent automaticity, given that activation of a
supporting goal did not facilitate habit performance.
The alternative position, that habits are a type of goal-dependent
automaticity, has been advanced through a creative series of stud-
ies by Aarts and colleagues on transportation and drinking habits.
Aarts and Dijksterhuis (2000, Study 1) found that college students
who frequently rode their bikes gave faster judgments about
whether the bike was a realistic means of transport to a number of
locations when they had been primed earlier with relevant trans-
portation goals (e.g., going to class) than when they had not been
primed with goals. In the researchers’ interpretation, cycling habits
were shown to be goal dependent because cycling-related judg-
ments were facilitated by goal activation. However, because the
paradigm was limited to judgments, the findings plausibly reflect
goal-activated explicit beliefs and judgments as opposed to proce-
dural memory associations underlying habit performance. Closer
to the study of actual habit responses is Sheeran, Aarts, et al.’s
(2005) demonstration that activating goals to socialize can prompt
students with a habit for alcohol consumption to sign a voucher for
a free drink. However, signing the voucher presumably was not an
element of participants’ typical drinking habits but instead was a
novel action that, again, likely involved reflection and decision-
making. Thus, in our view, the data do not show that participants’
actual drinking habits were goal dependent. It remains to be seen,
for example, whether activation of the socializing goal would
prompt participants with strong drinking habits, upon leaving the
experiment, to head for the bar at which they habitually drink.
In summary, different patterns of responding appear character-
istic of automated goal pursuit and habit automaticity. Perfor-
mance guided by implicit goals often depends on currently held
explicit goals, whereas habit performance does not do so. Implicit
goal pursuit that shifts with changes in people’s explicit goals
appears to yield variability in responding as opposed to the rigid
repetition typical of habitual responses.
Role of Goals in Habit Models Across Psychology
The idea that habit associations do not involve the mediation of
goals is supported by findings from diverse areas across psychol-
ogy. We briefly review here three relevant literatures: research on
the neural systems underpinning repeated responding, computa-
tional models of the cognitive processes that underlie routine
action, and animal learning studies of habitual and nonhabitual
In a typical neuroimaging study of habit formation, the neural
correlates of task performance are monitored as participants repeat
a motor task until it becomes habitual according to some behav-
ioral criterion (e.g., absence of dual-task interference effects).
Repetition and the consequent development of habitual control
typically are associated with a significant redistribution of brain
activity (see reviews in Jonides, 2004; Kelly & Garavan, 2005).
Note that the neural correlates of habit development usually fea-
ture reduced activation in the prefrontal cortex (PFC) and in-
creased reliance on subcortical structures including the basal gan-
glia and cerebellum. The PFC is considered critical to the selection
and pursuit of goals (e.g., Banfield, Wyland, Macrae, Mu¨nte, &
Heatherton, 2004; E. K. Miller & Cohen, 2001). Thus, evidence of
the progressive disengagement of the PFC during habit formation
is consistent with a shift away from goal-based control as re-
sponses are repeated into habits.
Also relevant to understanding the role of goals in habit perfor-
mance are schema-based and connectionist models that describe
the control of routine action. In theory, models that represent
routine action in terms of cognitive schemas assume that all action
is organized by goals (e.g., Cooper & Shallice, 2006; Norman &
Shallice, 1986), whereas models that represent routines in connec-
tionist networks eschew a goal-mediated structure for habitual
responses (e.g., Botvinick & Plaut, 2004, 2006). However, Botvin-
ick and Plaut (2006) argued that, in practice, even schema-based
models accord limited significance to goals in executing routine
responses. Schema model simulations represent a goal as a gating
mechanism or negative precondition rather than as “a representa-
tion of a desired outcome that is matched against action effects as
part of a process of means-ends analysis” (Botvinick & Plaut,
2006, p. 923). Within schema models, goals assume a means–ends
function only within alternative systems of action control designed
for planning nonroutine actions (e.g., Norman & Shallice’s, 1986,
supervisory attentional system). Thus, a range of computational
models effectively simulate habit-based responding with only lim-
ited representation of goals and action outcomes.
Animal learning research provides behavioral evidence for the
independence of habits from goal systems. In the reinforcer de-
valuation paradigm, when an animal has initially mastered an
instrumental response, such as a rat pressing a bar for a food pellet,
performance depends on goal-relevant outcomes (see Dickinson &
Balleine, 2002). For example, bar pressing will rapidly desist if the
food pellet reinforcer is devalued by feeding the rat to satiation or
by pairing the pellet with a toxin (e.g., Adams, 1982; Dickinson,
Balleine, Watt, Gonzalez, & Boakes, 1995). However, if the re-
sponse has been practiced to the point of being habitual, reinforcer
devaluation ceases to have an immediate impact on response
performance, suggesting that the response is not closely dependent
on the value of its outcomes.
Animal studies of place learning in mazes also suggest a re-
duced role of goals along with an increased role of context as
animals acquire habits (Packard, 1999; Packard & McGaugh,
1996). Initially, rats navigate through a maze as if guided by a
mental map of the location of the reward. Even if placed in a new
starting position, they successfully navigate toward the reward
location. However, with extended maze training, behavioral con-
trol shifts to be based on a series of S-R links that are triggered by
context without appearing to encode the ultimate reward destina-
tion. If placed in a new starting position, rats with extensive
training implement the context-cued turns appropriate for their
typical starting position instead of the ones that lead to the reward.
Thus, in animal paradigms of reward devaluation and place learn-
ing, habit performance appears to be guided by mental represen-
tations of associated contexts and responses rather than by the
rewards that initially spurred performance.
Despite the use of varied methods and theoretical frameworks,
neuroimaging research, computational modeling, and animal
Neuroimaging studies of the reduced engagement of the PFC with
habit development typically focus on conscious– explicit goals and do not
address nonconscious–implicit goal pursuit. Suggesting that these two
forms of goal pursuit rely on essentially the same neural structures,
Pessiglione et al. (2007) reported that supraliminally and subliminally
presented monetary cues engage essentially the same basal forebrain sys-
tems. This supports the idea that reduced PFC engagement with habit
formation signals reduced reliance on both conscious– explicit and
nonconscious–implicit goal pursuit.
Goal expectancies that develop with repetition provide a possible
alternative account of the insensitivity of extensively trained responses to
changes in reward value. That is, insensitivity could arise because stronger
expectancies are formed with extensive than moderate training, and stron-
ger expectancies are likely influenced less by episodes of devaluation.
Countering this alternative explanation is research showing that insensi-
tivity to reinforcer devaluation can be accelerated or delayed by lesions in
habit-related brain systems (Coutureau & Killcross, 2003; Yin, Knowlton,
& Balleine, 2004). Thus, habitual control, as indexed by insensitivity to
reinforcer devaluation, can be manipulated in part independently of rates of
repetition and presumably any associated repetition-based expectancies.
learning studies converge upon a common conclusion: The neural,
cognitive, and motivational substrates of action control appear to
shift with repetition of responding and habit development so that
performance is not mediated by goals.
Thus far, we have outlined how habits are represented in the
association between responses and recurring aspects of perfor-
mance contexts (Principle 1), and how these associations are not
mediated by goal representations (Principle 2). However, these
principles do not imply that habits necessarily are implemented in
a manner that is devoid of input from people’s goals. Instead, as
delineated by the third principle of our model, habits interface with
goal systems in certain limited ways.
Habits and Goals
The habit–goal interface is constrained by the particular manner
in which habits are learned and represented in memory. Specifi-
cally, the associative learning underlying habits is characterized by
the slow, incremental accrual of information over time in proce-
dural memory (see Graybiel, 1998; Packard & Knowlton, 2002;
although also see Pasupathy & Miller, 2005). The slow time course
of such learning is critical because it insulates habit dispositions
against short-term changes in behavior that occur as people flex-
ibly pursue their goals. Thus, habit dispositions undergo minimal
change to reflect current goals or occasional counterhabitual re-
sponses. Only with extended repetition in stable contexts are
behavior patterns likely to be represented in habit learning.
Slow-learning, conservative memory systems, as exemplified in
habits, appear to confer functional benefit for learning systems. By
reflecting the recurring features of an organism’s past experiences,
such systems shield existing knowledge against potential disrup-
tion from being overwritten or unduly distorted by new experience.
For example, in McClelland et al.’s (1995) connectionist model-
ing, the problem of catastrophic interference, by which newly
learned inputs significantly disrupt existing patterns of knowledge,
occurred in models that did not incorporate slow-learning systems.
With respect to habits, the functional benefits of insulating estab-
lished learning from the vagaries of changing goals have been
captured in Daw et al.’s (2005) computational model of reinforce-
ment learning. This model incorporates complementary roles for
habitual control versus goal-mediated (in their terms, cognitive,
goal-dependent) control. Habitual control is computationally sim-
ple; inflexible; and, as we explained in our description of moti-
vated cuing, oriented toward the cached, cumulative value of a
response. In contrast, goal-mediated control is computationally
expensive, flexible, and dynamically sensitive to changes in goal-
relevant outcomes. In Daw et al.’s dual system, each controller is
suited to guide responding in specific circumstances, suggesting a
unique functionality to conservative habit systems separate from
more flexible, goal-mediated control.
A critical implication of the representation of habits in slow-
learning neural and cognitive systems distinct from goals is that
habitual action control is not readily integrated with goal-based
control systems. Thus, when guiding action, habits and goals do
not form a unitary, averaged response disposition but instead
combine in various ways that have the effect of retaining the
integrity of the habit disposition. In the third principle, we outline
three ways that the slowly acquired nature of habit associations
constrains the interface between goals and habits.
Principle 3: Habits interface with goals.
Figure 1 depicts the ways in which habits and goals interface
according to the third principle of our model. In one form of this
No habit
Goal to
In contiguity
with cont
Habit learning:
No habit
Perceive cues
that activate
Figure 1. Illustration of the interface between habitual and goal-based systems of action control as outlined in
the third principle of the model. The left panel reflects how goals direct habit formation, the right panel reflects
the interaction between goals and habits in guiding performance, and the arrow at the bottom reflects how habits
inform goals, as when people infer goals and related dispositions from habit performance.
interface, illustrated in the left side of the figure, habit responses
operate in the service of goals. Typically, this form occurs because
goals direct control of responses prior to habit formation and thus
define the contexts under which a response is repeated into a habit.
Goals also direct habits when they lead people to encounter con-
texts that themselves activate habitual sequences of responding.
However, as we explain, the triggered habit may promote goal
pursuit or may subvert it (e.g., action slips).
A second way in which habits and goals work in concert to
guide performance occurs when past habit responding informs
people’s goals. This inference process is depicted in Figure 1 in the
arrow running from the bottom right to the bottom left. Habits are
informative in this way when people reflect on how they have
acted in the past in order to make inferences about their goals and
related dispositions, such as attitudes and personality traits.
Finally, the interface also allows for habits and goals to interact
in guiding action. This occurs when a habit disposition and goal
disposition are both available to control a given response. The
processes by which goals and habits interact are depicted in the
right side of Figure 1. Because the slow acquisition of context–
response associations precludes flexible integration of habits with
current goals, the interaction reflects that responding generally is
guided by one or the other disposition. Control of habit responding
in favor of conflicting goals involves effortful self-control to
override the automatically cued habitual response. Exerting such
control depends on available regulatory capacity to inhibit the
unwanted habit.
Habits in the Service of Goals
Goals spur habit learning. Goals guide habits most fundamen-
tally by providing the initial outcome-oriented impetus for re-
sponse repetition. In this sense, habits often are a vestige of past
goal pursuit. This is not to say that people’s habits always are in
line with their goals. By definition, unwanted or bad habits are in
conflict with goals, and furthermore action slips sometimes in-
volve the performance of an unwanted habit as opposed to an
intended response (Reason, 1990). However, given that habits
often originate in goal pursuit, the outcomes of habits should
generally accord with what people wish to achieve.
The correspondence between habits and desired outcomes was
demonstrated in Ouellette and Wood’s (1998) meta-analysis across
33 studies, which revealed that habit strength (as reflected in
frequency of past performance) was positively correlated with
favorability of behavioral goals (r .43, p .01). Furthermore, in
experience-sampling diary studies in which participants reported
each time they wanted to change some response, strong habits
proved to be a minority of the unwanted responses nominated for
change, with only about one fourth of those mentioned being
strongly habitual (Quinn, Pascoe, & Wood, 2007, Studies 1 and 2).
Although these correlational patterns are consistent with the idea
that habits originate in purposive action and hence tend to promote
desired ends, in the following section (Habits Inform Goals), we
address the possibility that such associations can reflect the reverse
causal order in which people infer goals from habits.
The goal-dependent process of habit learning can emerge
whether or not a person explicitly intends to tie responses to
context cues. Thus, in Figure 1, goals might direct habit learning
when people repeatedly implement goals to respond to a particular
context cue (e.g., skill learning, implementation intentions) as well
as when they repeatedly implement goals to respond that do not
specify contexts (e.g., implicit learning). In both cases, goals
provide initial top-down control until the response is practiced into
a habit and is cued by contexts in a bottom-up fashion.
1. Habit learning from goals to respond to a context cue.
Goals to respond repeatedly to a given cue underlie much of skill
acquisition, which involves people tailoring their responses to
environmental events with the goal of perfecting their performance
(e.g., feel dancing partner moving forward move back; see
Anderson, 1982). This phenomenon also is exemplified by imple-
mentation intentions, or plans to perform a response given a
specific cue (Gollwitzer, 1999; e.g., “I will put my fork down after
every bite”). By repeatedly enacting such behavioral goals in
stable contexts, people plausibly are slowly acquiring the context–
response associations by which responding can be cued without
goal mediation.
2. Habit learning from goals to respond. Habit learning also
can originate in goals when people are not aiming to respond to
any particular context cue. An extensive literature on implicit
learning shows that while purposefully repeating actions, people
learn associations between contexts and responses even when they
could not plausibly have any goal to learn them (e.g., Cleeremans,
Destrebecqz, & Boyer, 1998; Frensch & Ru¨nger, 2003). For ex-
ample, in sequential response tasks, people detect—apparently
without a goal or awareness—statistical relationships between
contextual stimuli and responses and then use this information to
guide responding. As illustrated in Lewicki, Hill, and Bizot’s
(1988) classic study, participants indicated as quickly as possible
when a stimulus shape appeared in one of the four quadrants of a
computer screen. Unknown to participants, the position of the
stimulus in some trials could be predicted by the order of preced-
ing stimuli. As participants gained experience with the task, the
speed and accuracy of their performance on predictable trials
improved relative to unpredictable trials. Thus, participants were
detecting the sequential context established by the preceding stim-
uli and were using it to facilitate their responding.
In implicit learning studies, participants hold global goals to
complete the experimental task as well as specific goals to perform
task procedures (e.g., press the buttons), but the facilitating effects
of specific cues (i.e., preceding stimuli) on responses cannot be
produced by an explicit goal given that participants typically
cannot report on the helpful response sequence (although see
Shanks & St. John, 1994; Wilkinson & Shanks, 2004). Further
suggesting that participants are not acting on a goal to respond to
the specific patterns of covariation, the effect appears robust to
manipulations of goals: Participants cannot refrain from express-
ing this learning during performance even when instructed to do so
(Destrebecqz & Cleeremans, 2001), and participants’ performance
fails to improve, at least on moderately complex sequences, when
they are instructed to discover the covariation pattern (Jime´nez,
Me´ndez, & Cleeremans, 1996). Thus, evidence of implicit se-
quence learning indicates that people incidentally can acquire
specific patterns of context–response associations in the course of
broader goal pursuit. The learned associations then can guide habit
Goals leading people into contexts that trigger habits. Habits
also function in the service of goals when, during goal pursuit,
people encounter context cues that trigger a habit. Habit perfor-
mance thus depends on a goal having led to the perception of
relevant context cues. For example, the first steps in getting ready
to drive to work in the morning may be too complex and variable
to be rendered completely habitual. A person may pursue goals to
organize the drive into work until he or she encounters a sequence
of stable context cues (e.g., get into the car). At that point, a
habitual sequence can be activated by the physical setting and by
preceding actions, thereby largely obviating the need for control by
goal systems. The habit component of this scenario is displayed in
the right panel of Figure 1. During goal pursuit that brings people
into a relevant context, the perception of relevant cues activates
habit performance that then proceeds in a goal-independent man-
When goal pursuit enables access to triggering habit cues, goals
are working hand in glove with habits—the former segueing into
the latter within an ongoing stream of action. The nature of this
transition is especially evident when the pursuit of a particular goal
leads people to encounter context cues that trigger a goal-
inconsistent habit. In the action slips literature, such events are
termed capture errors (Norman, 1981) or strong habit intrusions
(Reason, 1990). They are defined as highly automated action
sequences that are inadvertently triggered when a person intended
another (less habitual) response. In Reason’s (1992) diary studies
of everyday behavior, this kind of error constituted up to 40% of
all action slips. Such habit intrusions were especially common
when the habit shared “similar location, movements, and objects
with the intended actions” (Reason, 1992, p. 82). For example, a
person might set a goal to drive to the grocery store and, in doing
so, encounter context cues (e.g., the car, familiar streets) that have
become linked to his or her habitual drive to work. The cues might
trigger an inadvertent trip to that destination rather than to the
store. Such errors show how a goal state can work at the front end
of a sequence of actions that results in activation of a habit. That
the cued habit can be either consistent or inconsistent with the
initiating goal reflects the nature of habit cuing. Because the
initiating goal does not mediate habit operation, the habitual re-
sponse is activated whether it diverges from the person’s initial
aim or not.
In summary, habits and goals can promote each other in the
ongoing stream of everyday behavior. This interface is illustrated
by instances in which habits work in the service of goals. For
example, habits form as people repeatedly respond in contiguity
with context cues, either because they purposefully undertake to
give a response in a particular context or because they purposefully
respond in ways that happen to be contiguous with features of
performance contexts. In another example of the interface, habit
performance is indirectly guided by goals when goal pursuit leads
people to encounter contexts that activate a habitual response. In
these ways, habit performance can be considered a downstream
consequence of goal pursuit.
Habits Inform Goals
There is good reason to believe that the causal pathway between
goals and habits also can be reversed, so that people use their
habits to infer goals and other personal dispositions to value
particular response outcomes (e.g., attitudes, self-concept beliefs).
The process of using habits to infer subsequent goals and dispo-
sitions is depicted in the arrow at the bottom of Figure 1. As we
explain below, inferences of goals from habits are important be-
cause they may contribute to the regulation of habits with respect
to desired outcomes, although the habit itself is not mediated by a
goal representation.
Post hoc inferences from habits. The propensity to make in-
ferences from past behavior reflects basic properties of human
cognition. Specifically, people often have limited introspective
access to the causes of their thoughts and behavior (Nisbett &
Wilson, 1977). When internal causal states are weak, ambiguous,
or uninterpretable, people may be forced to draw inferences about
such states from their own behavior and other external cues (Bem,
1972). Furthermore, such inferences appear to be pervasive. Peo-
ple even infer humanlike dispositions from observations of the
behavior of animals and inanimate objects (Epley, Waytz, &
Cacioppo, 2007). The overall pattern is one in which people may
have fairly limited insight into the causes of particular overt
behaviors but nonetheless readily infer supporting goals and dis-
This inference process is represented in computational models
of routine action that allow for non-goal-mediated routine respond-
ing can give rise to goal representations. For example, in Sun,
Slusarz, and Terry’s (2005) CLARION model, habitual responses
that are controlled through bottom-up procedural knowledge can,
over time, come to be represented in top-down rules via a rule-
extraction-refinement algorithm. Similarly, Carver and Scheier
(2002) discussed the possibility that goals develop in a bottom-up
manner as emergent, self-organizing properties of dynamic sys-
tems rather than being imposed top down. As they explained,
“dynamic processes at a low level may automatically (without
intent) produce emergent patterns . . . with enough practice, the
patterned information can be used top-down” (p. 312).
Although people make inferences from habitual responses about
their goals and other dispositions, they are not completely insen-
sitive to the cued origins of such responses. Wood et al.’s (2002)
diary research compared the inferences participants made from
their habits with inferences from nonhabitual, presumably more
goal-dependent, forms of responding. In such comparisons, habits
were judged less a product of internal dispositions and relatively
uninformative to others about the self. The tendency to make less
certain self-inferences from habits as opposed to other behaviors is
consistent with evidence that procedurally based habit memory
contributes little to the behavioral autobiography on which the self
depends (for a review, see Klein, German, Cosmides, & Gabriel,
2004). This autobiographical self is thought to draw predominantly
from the episodic component of declarative memory, thereby
excluding procedurally based habits. Thus, people may underuti-
lize habits as a basis for self-inferences.
An alternative prediction, of substantial input from habits to the
self-concept, comes from Lieberman and Eisenberger’s (2004)
proposal that habits inform a distinct intuitive self that is not reliant
on the behavioral memories provided through episodic memory.
Supporting this idea, Lieberman, Jarcho, and Satpute (2004) re-
ported that people making self-judgments in domains in which
they had considerable experience, such as athletes judging them-
selves on athletic traits, showed activation of habitual control
systems (i.e., basal ganglia, ventromedial PFC, amygdala, and
lateral temporal cortex), whereas people making self-judgments in
domains in which they had little experience, such as actors judging
themselves on athletic traits, showed activation of nonhabitual
control systems (i.e., dorsolateral PFC and hippocampus). The
researchers argued that habits may inform a kind of implicit
self-knowledge that is not reliant on episodic memory. It remains
to be seen whether post hoc inferences of goals can be made on the
basis of this implicit self-knowledge.
It is thus unclear at present whether, as Lieberman and Eisen-
berger (2004) argued, habits are uniquely informative about peo-
ple’s intuitive self or whether, as Wood et al.’s (2002) diary
research and Klein et al.’s (2004) studies of episodic memory
suggest, habits are relatively uninformative about the self-concept
compared with other behaviors.
Self-regulatory implications of habits giving rise to goals. The
post hoc inferences that people make from their habitual responses
are important because they provide a potential mechanism through
which habits can be regulated in accordance with desired out-
comes, even though the habit itself is not goal mediated. Within
standard test– operate–test– exit models of self-regulation, goals
provide a desired comparison standard against which current cir-
cumstances are tested (Carver & Scheier, 1998; G. A. Miller,
Gallanter, & Pribram, 1960). When these tests indicate that be-
havioral outcomes deviate from those specified by a relevant goal,
people operate on the behavior, or exert self-control, to bring it in
line with the standard. In the next section of the article, we discuss
the means by which people exert self-control over habits.
In the case of habits, how could such testing processes unfold
given the lack of goal mediation? If people observe their habitual
behavior and impute plausible goals to those actions, then this post
hoc goal inference may provide a comparison standard by which
the habit can be regulated. The inferred goal may or may not be the
one that originally motivated response repetition and habit forma-
tion. For example, a student with a habit to do homework on the
computer after dinner may infer that the behavior reflects his or her
strong academic achievement goals. Once inferred, the goal can
then be used as a comparison standard in testing for goal-
inconsistent outcomes, such as surfing the Web. In this way, habits
may be regulated via standard test procedures involving compar-
ison with goals, even though a habitual response is not itself goal
Habits Interact With Goals to Guide Responses
A third dimension of the habit– goal interface arises when re-
sponses are habitual and yet also directly related to a currently held
goal state. In such cases, the slowly acquired context–response
associations do not merge with the goal but instead the two
dispositions interact in guiding behavior.
When habits and goals dictate the same response, our model
suggests that goals in effect are rendered epiphenomena, as action
control is outsourced to context cues that reliably covaried with
past performance. Behavior prediction research bears out such a
pattern in which goals typically correspond with, but do not appear
to guide, habitual responses. As we noted already, the standard
finding in behavior prediction research is that the strength of
people’s goals and the strength of their habits are each significant
predictors of future performance. However, when goals and habits
are considered jointly as predictors of future performance, they
typically interact in their effects. The interaction reflects that goals
do not predict future behavior when habits are moderate or high in
strength. Instead, performance is a product of the strength of those
habits. This interaction has emerged in predicting frequency of
driving a car, recycling waste, donating blood, watching TV,
exercising, voting in national elections, and purchasing fast food
(Aldrich, Montgomery, & Wood, 2007; Ferguson & Bibby, 2002;
Ji & Wood, in press; Ouellette & Wood, 1998, Study 2; Verplan-
ken et al., 1998; Wood et al., 2005). The interaction also has been
reported in a meta-analytic synthesis of condom use (Albarracı´n,
Kumkale, & Johnson, 2002) and a synthesis of studies assessing
various everyday behaviors (Ouellette & Wood, 1998, Study 1).
Also relevant, habit strength has been found to moderate the
effects of personal moral norms (Klo¨ckner, Matthies, & Hunecke,
2003) and self-concept as someone who performs the behavior
(Ouellette & Wood, 1998, Study 2, with respect to watching TV).
The findings conform to a pattern in which morals and self-
concepts cease to predict behavior at moderate and strong levels of
habit strength. Thus, behavior prediction research demonstrates
that people can hold goals that are consistent with, but do not
appear to guide, habitual responses.
Longitudinal evidence of the reduced role of goals with increas-
ing habit strength comes from Baldwin et al.’s (2006) investigation
of the determinants of quitting smoking among people voluntarily
participating in an 8-week smoking cessation program. At the
initial stages of quitting and during the early stages of maintenance
(i.e., quitting for 3 consecutive months), participants’ success was
predicted by aspects of their decision making and goals, including
perceived efficacy of quitting and satisfaction with the outcomes
of quitting. However, among the 13% of participants (n 74) who
had quit for 9 months after the end of the program, only the
number of continuous months that they had successfully quit
predicted whether they continued not to smoke at 15 months. As
Baldwin et al. (2006) explained, “once people have been quit for
a relatively long period of time, their behavior (i.e., not smoking)
becomes a function of itself (i.e., a habit) and, thus, is less sensitive
to variability in their beliefs” (p. 632).
This longitudinal design
corroborates cross-sectional behavior prediction work in demon-
strating that, once habits form, behavior is guided by the strength
of those habits, and goals become epiphenomena.
Habits and goals also have the potential to interact when the two
dispositions conflict in their implications for action. Because hab-
its represent the gradual accrual of context–response associations
that are represented in a non-goal-dependent form, habits remain
relatively intact in the face of new experiences and conflicting
Goals’ failure to predict future behavior when strong habits have
formed does not appear to be due to some weakness or uncertainty in the
goal judgments. Ji and Wood (in press) found that people with stronger
habits reported holding their intentions with greater certainty, even though
these intentions did not predict future behavior independently of habit
Readers may wonder whether nonresponses, such as quitting smoking,
can be considered habits. Many nonresponses do not fit the definition of
habits because they do not represent any particular learned association
between context cues and responses. However, nonresponses might pos-
sibly be considered habits when they reflect the formation of automatic
associations between cues and repeated acts of response inhibition or
between cues and alternative responses (e.g., chewing gum in response to
a cue linked to one’s smoking habit). Such cue–response associations can
be likened to those in studies addressing the learned extinction of goal-
directed behaviors (see Bouton, 2000).
current goals. Evidence that habits persevere even when in conflict
with goals comes from Webb and Sheeran’s (2006) synthesis of 47
studies using persuasive appeals and other interventions to change
people’s behavioral goals. The central question was whether the
interventions, which were selected because they significantly
changed behavioral goals, would also change behavior. The an-
swer depended on the habit strength of the behavior. Interventions
that addressed behaviors conducive to habit formation, in that they
could be performed frequently in stable contexts (e.g., seat belt
use), yielded only minimal behavior change (d 0.22, k 35).
However, interventions that addressed behaviors that were not
conducive to habit formation (e.g., course enrollment) yielded
more substantial behavior change (d 0.74, k 12). Thus, when
people could form habits, they continued to perform the habitual
response despite having adopted new behavioral goals.
Behavior prediction and intervention studies thus converge in
demonstrating the moderating impact of habit strength on the
predictive power of goals. Once habits are formed, goals are either
epiphenomena (when in concert with habits) or exert little mod-
erating impact on actual behavior (when in conflict with habits).
Even though goals appear to have limited influence when in
conflict with established habits, people obviously can exert self-
control over many of their goal-inconsistent habitual responses. It
is simply that with the behaviors studied in behavior prediction and
intervention research, people typically do not do so. Limited
self-control over everyday habits also is plausibly evident in self-
regulatory failures that involve the repetition of unhealthful or
otherwise unwanted behaviors (e.g., alcoholism, overeating, drug
addiction). In these failures, people often are aware to some extent
that their behavior deviates from desired standards. Baumeister
and Heatherton (1996) thus concluded that people are in some
sense complicit in many regulatory failures. As we explain, in the
case of habits, complicity involves failing to operate, or exert
self-control, over the offending habit so that it is inhibited in line
with a conflicting goal.
Self-control over habits as opposed to other cued responses.
One challenge to regulating habits is that they do not merge readily
with conflicting goals, and therefore habit dispositions are not
changed simply by adopting new goals or engaging in short-term
behavior change. Instead, the means of effective regulation come
from control over habit cuing.
Controlling stimuli and responses to them is central to regula-
tion of other types of cued responses, including visceral reactions
(e.g., Loewenstein, 1996) and impulsive hot responses (Metcalfe
& Mischel, 1999). However, the specific forms of such control that
gain traction over affective and visceral responses are not neces-
sarily the same as those that gain traction over habits. People can
control visceral or emotional responses triggered by a stimulus by
cognitively minimizing the affective qualities of cues. For exam-
ple, in Metcalfe and Mischel’s (1999) research on delay of grati-
fication, children were better able to delay gratification when
actively reinterpreting a tempting food treat in a manner designed
to reduce its affective qualities, such as likening marshmallows to
clouds, or when distracting themselves, perhaps by thinking about
something else.
The cognitive strategies useful in controlling affectively based
responses generally will be less successful with the direct cuing of
habits. Given that habitual responses are directly activated by
perception of cues, control of habits is not likely facilitated by
altering affective properties of those cues. In fact, being distracted
or otherwise preoccupied appears to promote the performance of
unwanted habits, as evidenced in Reason’s (1990, 1992; see also
Botvinick & Bylsma, 2005) work on action slips in everyday life.
Furthermore, as we explain below, instead of inattention to the cue,
high levels of vigilance to it appear to be effective at self-control
of unwanted habit responses.
In contrast to the cognitive methods used to control affective
cuing of impulsive responses, the directly cued nature of habit
responding is sensitive to two particular forms of cue control.
These are (a) inhibiting the performance of the habitually cued
response once it has been activated and (b) altering actual exposure
to the cue so as to avoid initial triggering of that response. This
latter strategy is likely to be an all-purpose means of cue control
that works with habits as well as other cued responses (see Met-
calfe & Mischel, 1999).
Inhibiting the performance of cued habit responses. One way
for people to control habit cuing is through sheer dint of will. That
is, people may implement effortful control to override the habit
disposition and prevent it from manifesting in behavior.
1. Effortful self-control to inhibit habits. The capacity to in-
hibit habits appears to depend critically on people’s dynamic levels
of self-control. Self-control can be considered a finite, domain-
general resource that is depleted when people attempt effortfully to
inhibit or override thoughts, emotions, and behaviors (e.g., Mu-
raven & Baumeister, 2000). From this perspective, inhibiting hab-
its requires sufficient regulatory capacity.
Demonstrating the relation between self-control capacity and
habit inhibition, Vohs, Baumeister, and Ciarocco (2005) found that
participants who first engaged in a resource-depleting task, such as
a Stroop color-naming task, subsequently were less able to over-
ride characteristic (i.e., habitual) self-presentations in interactions
with others. This research also demonstrated that overriding ha-
bitual self-presentations, such as presenting oneself as having
gender-inconsistent attributes, reduced ability to self-regulate in
subsequent tasks that require self-control, such as physical stamina
in maintaining a hand grip (see also Tice, Butler, Muraven, &
Stillwell, 1995). Thus, fluctuations in self-control resources appear
to impair the inhibition of habits, and conversely, the inhibition of
habits appears to deplete self-control.
More directly relevant to real-world habits, Neal and Wood
(2007) conducted a daily diary study to investigate whether the
impact of self-control depletion on habitual and nonhabitual be-
havior can impair both elements of successful self-regulation.
Students identified a set of behaviors that they currently were
attempting to implement (e.g., getting to class on time) and a set of
behaviors that they were attempting to inhibit (drinking alcohol on
weeknights), and their performance of these was monitored over a
4-day period. For 2 of the 4 study days, students’ self-control was
reduced by requiring them to use their nondominant hand for a
range of everyday activities, thereby imposing a sustained inhib-
itory drain. On the days when self-control was lowered in com-
parison to when it was not, participants were significantly more
likely to fail at inhibiting habitual behaviors. In contrast, self-
control depletion had minimal impact on behaviors that partici-
pants wanted to implement as well as behaviors that they wanted
to inhibit that were not habitual. Thus, comapred with the control
of other behaviors, the inhibition of strong habits appears to
depend upon the availability of sufficient self-control capacity.
Thus, self-control depletion appears to impair two facets of self-
control, involving the inhibition of existing habits and the imple-
mentation of new responses in place of familiar habits. These
findings highlight the various ways that self-control capacity is a
limiting factor in regulatory efforts to change habits.
The inhibition of unwanted habits may be aided by the use of
particular approaches to self-control. An avoidance strategy is
especially suited to inhibition given that it involves monitoring for
exposure to triggering cues, vigilant control to ensure that the
response is not elicited, and focus on the negative outcomes of
performing a habitual response (see Fo¨rster, Higgins, & Bianco,
2003; Freitas, Liberman, & Higgins, 2002).
Demonstrating the utility of avoidance in the context of inhib-
iting unwanted behaviors, Quinn et al. (2007, Study 2) conducted
an experience-sampling diary study of people’s everyday attempts
to change their responding. Participants reported each time they
wished to change a thought, feeling, or behavior as well as the
strategies they used to do so. Several days later, they reviewed
their reports and rated the ultimate success of each change attempt.
When change involved inhibiting a response rather than initiating
one, participants were especially likely to report using avoidance-
type strategies. Even more important, participants reported being
more successful at inhibiting unwanted responses when using
avoidance strategies rather than approach strategies that involved
focusing on the desired response.
Quinn et al. (2007, Study 3) also conducted a follow-up exper-
iment to clarify the causal relations suggested by the diary re-
search. The experiment modeled the inhibition of habits using a
laboratory word-association task (following Hay & Jacoby, 1996).
In this task, participants’ habit-based memories can be manipu-
lated to be either in concert or in conflict with their intention-based
memories. Replicating the findings from the diary study, Quinn et
al. found that participants were most successful at inhibiting con-
flicting habits when instructed to use an avoidance strategy of
being vigilant for errors and trying not to make mistakes by
responding habitually to the memory cue. That is, avoidance was
more successful than the approach strategy of striving to perform
well and than a control condition in which participants were given
no instructions. Furthermore, calculations of the amount of control
exerted over responding indicated that an avoidance strategy dis-
rupted habit performance by increasing successful exertion of
control over responding rather than by reducing the influence of
habit (see Hay & Jacoby, 1996).
It remains to be seen precisely how avoidance strategies en-
hanced habit inhibition in Quinn et al.’s (2007) diary and experi-
mental data. One possibility is that avoidance increases the like-
lihood that people recruit the counterhabitual goal rather than
recruiting only the habitual response. Another possibility is that
avoidance somehow increases inhibitory capacity or sharpens the
efficiency of inhibitory efforts so as to reduce their regulatory
In summary, self-regulation of habits to align with goals can
proceed through effortful inhibition of the cued response. This
pattern reflects that goals gain little traction by themselves on the
slowly accrued context–response associations that make up habits.
Additional evidence of the effortful inhibition required for goals to
control habit cuing comes from research demonstrating that habits
are not readily regulated via automated behavioral goals that
counter the habitual response.
2. Inhibition via automatically activated goals? In several
studies, participants have automated goals that conflict with habits
by forming implementation intentions linking context cues and
habit-inconsistent responses (e.g., Betsch, Haberstroh, Molter, &
Glo¨ckner, 2004; Holland, Aarts, & Langendam, 2006; Verplanken
& Faes, 1999). Although this enables the cue– goal association to
be activated alongside the habit response, in our perspective, the
automatic goal should have limited impact in breaking or changing
the habit.
In evidence that automatic goals cannot break habits in this way,
Betsch et al. (2004) established habits in a transportation game in
which participants took certain subway routes to a final destination
(see similar findings by Verplanken & Faes, 1999). After initial
practice trials, Betsch et al. switched the correct routine, and
participants were told to take alternative routes. Even though
participants formed implementation intentions, so that they auto-
mated their new intentions by linking new routes to cuing events
(e.g., to go to A-town, take blue line), they erred on about 50% of
their responses by giving the previously practiced routes. Further-
more, participants’ errors occurred despite the fact that the coun-
terhabitual implementation intentions were reinforced by perfor-
mance-contingent payment.
The alternative conclusion, that automatic intentions can break
habits, was drawn by Holland et al. (2006) in a study of habits to
dispose of plastic cups and paper. Across all participants, forming
implementation intentions to recycle decreased the amount of trash
thrown in the regular waste bins, thus apparently breaking habits of
trash disposal. However, the analyses did not report the efficacy of
implementation intentions as a function of participants’ initial
trash-disposal habits. Thus, it is unclear whether this intervention
succeeded in changing behavior for those with strong habits to
throw trash in the waste bins. We would anticipate that automatic
goals had traction primarily over weak habits. Change of strong
habits should require sufficient regulatory resources to inhibit the
habitual response and implement the goal-consistent one.
Our proposal that habits can be regulated by control of cuing
highlights a second set of regulatory mechanisms to promote goal
pursuit in the face of conflicting habits. That is, control can arise
from altering exposure to context cues so as to avoid initial
triggering of associated habitual responses. Such a strategy is
reminiscent of Metcalfe and Mischel’s (1999) analysis of delay of
gratification in which children inhibit impulsive responses by
reducing the salience of hot stimulus cues (e.g., obscuring a
tempting food treat out of sight).
Altering exposure to context cues. People’s narrative accounts
of their own change attempts suggest the usefulness of altering
exposure to cues in the performance environment. In Heatherton
and Nichols’s (1994) research on everyday behavior change, ap-
proximately 36% of participants’ reports of successful change
attempts involved moving to a new location, whereas only 13% of
reports of unsuccessful attempts involved moving. Also, 13% of
successful change reports involved some alteration in the imme-
diate performance environment, whereas none of the unsuccessful
reports involved such shifts in cues.
Of more direct relevance to habits, behavior modification ap-
proaches have long recognized the benefits of altering perfor-
mance contexts in order to disrupt habit cuing. For example,
stimulus control is a component of addiction treatments in which
addicts are trained to avoid situational triggers such as the loca-
tions of past use and the presence of other users (e.g., Witkiewitz
& Marlatt, 2004). Typically, therapeutic habit change interventions
rely on people’s effortful attempts to limit their exposure by
altering or avoiding habit cues in their environment. Such attempts
to alter cue exposure can themselves require exerting some level of
self-control. After all, placing groceries out of sight in the kitchen
may successfully reduce consumption by altering food cues (Sobal
& Wansink, 2007), but it might require some effort to remember to
do so and to inhibit snacking while placing them there.
An alternative to intentional control over exposure to habit cues
arises with naturalistic changes in life circumstances that alter the
contexts in which people perform everyday habits. To illustrate
this possibility, Wood et al. (2005) studied college students trans-
ferring to a new university. One month before and 1 month after
the transfer, students were contacted to report on several everyday
behaviors (i.e., exercising, reading the paper, watching TV). Some
of the students reported that the transfer brought about changes in
the performance context for these activities—including changes in
locations (e.g., exercising at the gym) and interaction partners
(e.g., reading the newspaper with one’s roommate). Whether par-
ticipants maintained habits for performing these behaviors at the
new university depended on the consistency of the performance
context. Participants with strong habits at the old university who
reported that the performance context did not change across the
transfer also maintained their habits. For example, a regular pattern
of reading the paper at the old university carried over to the new
university. Furthermore, the carryover occurred regardless of stu-
dents’ behavioral goals for reading the paper at the new university.
However, participants with strong habits at the old university who
reported that features of the performance context changed with the
transfer did not maintain their habits. With a change in context,
students apparently were spurred to think about their behavior, and
despite their old habits, their actions came in line with their goals
at the new university. In contrast, context changes did not matter
for students with weaker habits; they acted on their goals both
before and after the transfer. Additional analyses revealed that for
those with strong habits, changes in behavior with the transfer
could not be explained through the new contexts producing
changes in goals (see Wood et al., 2005).
In the naturalistic changes in life circumstances evaluated by
Wood et al. (2005), people’s exposure to habit cues was altered by
external forces that did not require regulatory efforts either with
respect to avoiding triggering cues or with respect to inhibiting
responses once cued. The serendipitous change in context ap-
peared to liberate responses so as to be sensitive to goals, as
evidenced by the influence of goals in guiding performance.
In summary, when the outcomes of habitual responding conflict
with outcomes that people wish to obtain, the slowly acquired
context–response learning underlying habit dispositions does not
shift readily in accord with people’s current goals. Instead, habits
and goals interact such that one or the other guides responding. We
identified two mechanisms through which people regulate un-
wanted habits in ways that promote goal pursuit. First, habits can
be controlled through effortful inhibition of performance once
triggered. Habit change as an inhibitory process depends on the
availability of sufficient self-control resources. Second, habits can
be controlled through altering exposure to the cues themselves, and
altering cues sometimes may require self-control resources.
Through these two regulatory mechanisms, habit cuing is disrupted
so as to bring behavior in line with goals. We discuss these points
further in the next section of the article, with respect to habit
change interventions.
Summary and New Directions From the Habit Model
We presented our model in the form of three principles. Spe-
cifically, habits are a form of slowly accrued automaticity that
involves the direct association between a context and a response
(Principle 1), so that the context can activate the response without
the mediating involvement of a goal (Principle 2). Furthermore,
habit development and performance interface with the purposive
dimension of mental life as represented in people’s goals (Princi-
ple 3).
The advantage of conceptualizing habits in this way is evident in
the range of empirical findings accounted for by our model. These
can be summarized in a core pattern in which the slowly accrued
context–response associations, once established, guide perfor-
mance repetition without depending on people’s current goals.
This pattern plays out in various ways in people’s overt responses.
Specifically, explicitly held goals to respond appear relatively
unsuccessful at predicting subsequent habit performance (e.g.,
Ouellette & Wood, 1998). Established habits also maintain despite
changes in people’s goals to respond that are held explicitly (Webb
& Sheeran, 2006) or that are automated through planning (Betsch
et al., 2004). The habit pattern also is evident in neuroimaging data
showing reduced reliance on goal-related brain systems during
habit performance (e.g., E. K. Miller & Cohen, 2001). It also
emerges in animal learning paradigms in which habit performance
persists despite changes in goal-relevant outcomes (i.e., reinforcer
devaluation studies, Dickinson & Balleine, 2002) or changes in the
specific responses required to achieve goal outcomes (i.e., place
learning studies, Packard, 1999). In these ways, habit dispositions
are relatively insulated from the effects of adopting and pursuing
new goals.
The utility of our conceptualization of habits also is evident with
respect to Principle 3’s articulation of the multiple ways in which
habits can interface with goals. This interface takes particular
forms that are constrained by the slowly accruing nature of
context–response associations. First, habits can work in the service
of goals. Consistent with the idea that people can form habits when
they repeatedly pursue a particular means to a goal in a given
context, habits typically remain correlated with and thus continue
to serve people’s goals (e.g., Ouellette & Wood, 1998). We also
speculated that people can, through goal pursuit, place themselves
in contexts that cue habits. Second, people can infer goals from
their habitual behavior, and they plausibly use these post hoc
inferences in self-regulatory processes to guide habit responding.
Third, goals and habits interact when both are present to guide
performance. Specifically, when in concert with habits, goals tend
to be epiphenomena in guiding behavior (e.g., Ouellette & Wood,
1998). When in conflict with habits, goals by themselves have
limited capacity to break habits, except when alterations occur in
the cues that trigger habits (Wood et al., 2005) and when people
exert effortful self-control to inhibit habit performance and, when
desired, to implement new, goal-consistent behaviors (e.g., Neal &
Wood, 2007; Quinn et al., 2007; Vohs et al., 2005).
Forms of the Habit–Goal Interface
Our model provides a framework to generate new research
questions concerning the ways in which habits and goals can
interface in guiding action. Illustrating these new areas of inquiry
is the variety of ways in which habits can act in the service of
goals. For example, habitual context–response associations, hav-
ing become decoupled from the originating goal, may be open to
be co-opted in pursuit of alternative goals. That is, habits can come
under the top-down control of new goals unrelated to those that
initiated habit formation. Such co-opting might occur because
there is no scope for interference between a new goal and the
non-goal-mediated habit disposition.
Experimental findings suggestive of co-opting of habits by goals
come from task-switching research using a simple key-pressing
skill acquisition task (Mayr & Bryck, 2005). This paradigm is ideal
to reveal co-opting of habits because each cued press response
could potentially meet multiple goals. Specifically, to indicate
what computer key to press on a given trial, participants were
given a rule or goal indicating movement across the keypad in a
particular direction (e.g., clockwise, vertical). When participants
had only a little practice, switching a rule across trials inhibited
speed of responding. That is, participants responded more slowly
to the rule to move clockwise (e.g., from the top right of the
keypad to the bottom right) if the previous trial had involved the
same key-press movement (e.g., from the top right to the bottom
right) to the rule to move vertically. Goals for action were thus
sticky, and responding in a given manner for one goal interfered
with subsequent responding in the same manner for a different
When respondents were given extensive practice at Mayr and
Bryck’s (2005) key-pressing task, their responding became habit-
ual, and the goals no longer stuck to the cue–response association.
For example, when participants did (versus did not) extensively
practice a response under the vertical rule, their performance was
facilitated when switching to using the response under a clockwise
rule. Having just made a response enabled participants to make it
quickly again, despite that the two responses were in service of
different rules (goals). Although it is admittedly speculative, the
idea that new goals can co-opt habits that produce a goal-relevant
outcome follows from the lack of goal-mediation of the habit
disposition. Having become decoupled from goal systems, habits
do not suffer interference due to the original goal that formed the
habit association. It also may be that this phenomenon contributes
to instances of skill transfer (Barnett & Ceci, 2002). We suggest
that this is just one of many unexplored ways that habits interface
with goals while retaining the basic structure of context-cued
Regulation of Habits
Scholars historically have questioned the logic of habit perfor-
mance without input from goals, because it is not clear how people
would regulate such responses. Humans do not persistently per-
form habits entirely without regard to response outcomes, and thus
habits must be subject to some form of regulatory processing. As
G. A. Miller et al. (1960) argued, traditional S-R models fail to
accommodate regulation because they lack a feedback process for
determining whether actions are moving toward or away from
some goal or desired outcome. In our new model, habits are
regulated in part through their interface with goals. As we ex-
plained in Principle 3, even though goals may not be required to
mediate habit performance, people can infer in a post hoc manner
what goals might be served by their habits. We suggest that this
inferred goal can then serve as the comparison standard posited by
traditional test– operate–test– exit models, enabling tests for dis-
crepancies between the inferred goal and the actual habit out-
comes. People then can operate on the habit to better align re-
sponding with the inferred goal.
The process of matching behavioral outcomes to inferred goals
may be sufficient to account for many instances of habit regulation
of habits. Nonetheless, it is also possible that some basic forms of
regulation also can proceed entirely in the absence of goals. A
plausible regulatory mechanism for habits and other forms of
response that do not depend on goals is emerging from work on the
neural basis of conflict detection. Apparently, people can identify
errors in diverse forms of responding without a representation of
the correct response or any feedback regarding the outcome of the
action or representation of the correct, desired response (Botvin-
ick, Braaver, Barch, Carter, & Cohen, 2001; Yeung, Botvinick, &
Cohen, 2004). This capacity is thought to be subserved by the
anterior cingulate cortex (ACC), which is an executive PFC circuit
located on the medial surface of the frontal lobes.
Activity in the ACC consistently increases following errors in
choice reaction time tasks (e.g., Gehring, Goss, Coles, Meyer, &
Donchin, 1993). Initially, this activation was thought to show that
the ACC is responsible for detection of errors or mismatches
between actual responses and intended, correct responses. More
recently, imaging data and connectionist simulations have sug-
gested that ACC activation during performance of these tasks is
not signaling the detection of errors per se but rather the presence
of conflicts between multiple activated responses (see Botvinick et
al., 2001; Yeung et al., 2004). One consequence of ACC activation
is reengagement of the PFC and more purposive, conscious guid-
ance of action to address the conflicting response tendencies.
Specifically, when the ACC detects the presence of multiple com-
peting responses or the absence of any clear response option, it can
signal an imminent error without directly assessing response ac-
curacy or registering feedback regarding actual outcomes. Con-
scious control via goal-mediated systems can then be engaged to
guide behavior in line with current goals.
With respect to habits, conflict detection mechanisms may prove
relevant when multiple responses are cued by a given context, as
when people’s automated goals activate one response and habit
dispositions another. In such cases, conflict detection mechanisms
may signal the need for controlled processing simply by detecting
the presence of two incompatible responses. Although it remains to
be seen whether response conflict constitutes a broadly applicable
mechanism of habit regulation, it has the potential to liberate
theorizing about regulatory control from assumptions about com-
parison standards in the form of goals.
Habits and Implicit Goal Pursuit
An additional question raised by the present framework con-
cerns the conditions under which response repetition leads to habit
formation. To this point, we have been silent about the factors that
determine whether repetition leads to habit formation versus other
forms of automaticity, especially automatic goal pursuit. Although
we can only speculate on these processes, it seems plausible that
any factors that ensure continued activation of a goal during
learning of context–response associations will promote goal-
dependent automaticity over habits. When goals remain active
during the development of automaticity, the associative structures
that form through repetition are plausibly more likely to continue
to incorporate goals rather than direct context–response associa-
tions as reflected by habits.
What factors might promote the continued activation of goals in
automated responding and thus undermine the transition to habit-
ual forms of automaticity? Response complexity is one possibility,
given that complex responses are likely to require continued ref-
erence to the goal to ensure effective performance. Our reasoning
here draws on animal learning research that has explored the
moderating effect of response complexity on habit development. In
this literature, complex tasks are operationalized as ones in which
an animal may “execute either of two different actions to obtain
two different rewards” (Daw et al., 2005, p. 1705). In support,
behavioral data suggest that complex responses continue to be
sensitive to goal value, even given extensive training and hence
opportunity for habit formation (Colwill & Rescorla, 1988).
Another possible factor that might preserve goal activation,
thereby hindering habit development, is the extent to which con-
text cues are associated with few, rather than many, responses.
This prediction builds off of the idea of goal equifinality, which
reflects the extent to which a given goal is linked to multiple
possible behavioral means (captured in the expression, “All roads
lead to Rome”). In goal systems theory, the link between a goal
and any one behavioral means is diluted in proportion to the
number of other means to which the goal is linked (Kruglanski et
al., 2002). Similarly, we suspect that the greater the numbers of
behaviors linked to a given context, the lesser the capacity for the
context to cue any one behavior directly. For example, cues such
as one’s mother are associated with a number of different re-
sponses, potentially yielding conflict in responding that could be
resolved by consulting relevant goals. Thus, the effects of cues
associated with multiple responses are not plausibly explained
through direct cuing and more likely are due to priming broad
goals (e.g., Shah & Kruglanski, 2002).
We speculate that attention to action is another factor that
promotes activation of goals despite continued repetition. Neuro-
imaging data suggest that automatic responses that receive atten-
tion during execution are likely to engage systems involved in goal
pursuit (i.e., PFC), even though the same automatic responses
executed without attention fail to engage such systems (Jueptner et
al., 1997; Rowe, Friston, Frackowiak, & Passingham, 2002).
Therefore, responses that attract continued attention may, with
repeated performance, be automated in ways that reflect this con-
tinued engagement of goals. Such conditions are likely to promote
the formation of automatic goal pursuit as opposed to habits. The
neuroimaging research on attention to automated action also has
implications for broader understanding of automaticity (Neal &
Wood, in press). Research paradigms that require participants to
attend to what would otherwise be unattended responses could
inflate the apparent goal dependence of the automated response
and, in turn, underestimate the incidence of habits and other
non-goal-dependent forms of automaticity.
Only indirect evidence supports our speculations regarding the
factors that promote automated goal pursuit as opposed to habits.
Especially given that our reasoning has drawn predominantly from
animal learning research and fairly low-level behavioral tasks,
systematic investigation is needed to test our suggestion that
continued activation of goals during repetition hinders habit learn-
ing and to determine whether the specific factors of response
complexity, equifinality, and attention to action promote this con-
tinued activation.
Interventions for Habit Change
Although our model broadly addresses the psychological mech-
anisms underlying habit formation and performance, we suspect
that its principles will be tested most extensively within the spe-
cific context of changing unwanted habits. Understanding how to
design successful interventions to bring about changes in habits is
of sharp interest, especially to clinical, health, and consumer
Our model offers a fresh approach to behavior change interven-
tions by highlighting the mechanisms through which effective
habit change is likely to be accomplished, in particular, by the
control of habit cuing. Although systematic behavior change in-
terventions undoubtedly involve a host of considerations in addi-
tion to cue control, we propose that this is a necessary component
of the habit element of behavior change. In this spirit, we identify
promising directions in control of habit cuing that could be elab-
orated into formal interventions to change habits.
Our approach to habit change contributes to the emerging
social– cognitive– behaviorist synthesis within psychology (see
Metcalfe & Mischel, 1999). In classic behavior modification ap-
proaches, behavior change is instigated largely through manipula-
tions of environmental contingencies (see Follette & Hayes, 2001).
For example, treatment of addiction might include stimulus control
through removing or avoiding people, places, and other stimuli
that in the past have been associated with the reinforcing value of
the addictive substance. With the cognitive revolution and shift to
an internal causal locus for behavior, the focus of change inter-
ventions shifted accordingly. Following this tradition, many cur-
rent models focus on changing people’s decision making about
their actions (e.g., Simonson, 2005) or their beliefs about action
outcomes and performance efficacy (e.g., expectancy–value mod-
els, Albarracı´n et al., 2003; health belief model, Glanz, Rimer, &
Lewis, 2002; protection motivation theory, Floyd, Prentice-Dunn,
& Rogers, 2000). Furthermore, the most popular model of habit
change in clinical and health settings, the transtheoretical model
(Prochaska, DiClemente, & Norcross, 1992), focuses broadly on a
range of experiential and behavioral processes through which
people can accomplish change in unwanted behaviors (e.g., self-
reinforcement, social support), without giving any special priority
to controlling cues in the performance context. According to our
synthetic model, people can break habits by exerting control down-
stream of a habit cue, after exposure to the cue has activated the
response in memory. Such control is exemplified by effortful
inhibition or suppression of the habitual response. Control also can
occur upstream of the cue, before the response has been activated.
Such control arises from decisions to avoid or alter the cue itself
(e.g., reducing the habitual reading of new e-mail by disabling the
autonotify option) or from exploiting naturally occurring changes
in cues (e.g., as when changing jobs or moving houses).
Control of habit cuing that is initiated downstream involves
actively inhibiting the cued response. Such inhibition appears to be
effortful and to draw on a limited regulatory resource (Neal &
Wood, 2007; Vohs et al., 2005). In our research, inhibition was
successful at overriding habit cuing especially when it took the
form of avoidance involving vigilant monitoring for the unwanted
automated response (Quinn et al., 2007). Avoidance focus appears
to enhance controlled processes and thereby the effectiveness of
inhibitory control.
Avoidance is likely just one of a number of strategies that are
effective at controlling habitual responses triggered by context
cues. Other possibilities are suggested by behavior modification
approaches, including counterconditioning or training to associate
the triggering cue with a response that is incompatible and thereby
conflicts with the unwanted habit. Although in naturalistic studies,
performing an incompatible behavior did not appear to be an
especially successful strategy for inhibiting unwanted responses
(Quinn et al., 2007), it has proved useful in more structured change
interventions (see Follette & Hayes, 2001).
Effortful inhibition of responding is known to have a number of
undesirable effects, and intervention strategies built simply on
such inhibition are unlikely to be sufficient to bring about long-
term change in habits. Although inhibition in our diary research
was successful at changing everyday responses in the short run
(e.g., Quinn et al., 2007), long-term inhibition has been found to
increase negative affect, generate preoccupied thinking about the
inhibited response (Polivy, 1998), and produce ironic effects in-
volving increases in the unwanted responding (Wenzlaff & Weg-
ner, 2000). In addition, it is unclear whether people can sustain
effortful inhibitory efforts in daily life. People’s capacity to inhibit
is reduced with everyday fluctuations in their self-control re-
sources (Neal & Wood, 2007), and people have difficulty sustain-
ing attempts to inhibit tempting behaviors that provide immediate
pleasure despite being inconsistent with longer term goals (see
Baumeister & Heatherton, 1996, on self-regulatory failures of
overeating, alcoholism, etc.). Although discussion of the mecha-
nisms by which people bolster their change efforts is beyond the
scope of the present article, people potentially can use a variety of
counteractive control strategies to foster adherence to long-term
goals in favor of immediate temptations (e.g., self-imposed pen-
alties, see Fishbach & Trope, 2005).
We speculate that effortful inhibition contributes most produc-
tively to behavior change interventions when the suppression of
habit performance is paired with learning and performing a new,
desired response. That is, inhibition might be effective as a short-
term strategy to suppress a habitual response so as to enable a new,
goal-consistent pattern of responding to be established. When the
new response is repeated in contiguity with context cues, new
habits might be formed. For example, a dieter’s effortful inhibition
of his or her unhealthy eating habits may promote long-term
behavior change only insofar as it creates a temporary window of
opportunity in which to establish new, healthful eating patterns. In this
view, the inhibition of habit cuing is a short-term means of control
that, although perhaps unsustainable, enables the development of new,
more desired patterns of response. However, we note that when newly
learned associations override older ones (e.g., extinction), the new
learning is inherently unstable such that the original learning may
readily recur under a variety of circumstances (Bouton, 2000; Schma-
juk, Larrauri, & LaBar, 2007).
Control of habit cuing occurs upstream when people’s exposure
to relevant context cues is altered or disrupted in some way. This
control of cue exposure can result from deliberate decision making
or from serendipitous changes in the habit performance context.
With respect to deliberate decision making, interventions that
promote avoiding contact with habit cues are widely used in the
treatment of addictions (e.g., Witkiewitz & Marlatt, 2004). Also
relevant to individual decision making, interventions that alter
simple cues in eating contexts have been found successful in
control of habits to overeat (Sobal & Wansink, 2007). For exam-
ple, the amounts of food and drink that people serve and consume
decrease with smaller sizes of plates, spoons, and glasses. Thus, by
using small plates and utensils, people could take advantage of the
fact that their habits for serving size are cued in proportion to
container size, with only limited adjustment for the absolute sizes
of the containers.
Individual efforts to control habit cuing upstream of a behavior
potentially suffer the drawbacks of lack of sustainability, preoc-
cupied thinking, and ironic effects found with active inhibition.
That is, attempts to alter or avoid exposure to habit cues may not
reduce the need for people’s effortful control but simply relocate it
to be needed earlier in the behavior stream. Altering or avoiding
cuing in some cases involves sustained effortful self-control to
identify the cues correctly and to avoid exposure to them. Like
effortful inhibitory strategies, this effortful control of cue exposure
is likely vulnerable to fluctuations in self-control resources and to
ironic and other counterproductive effects that derail change ef-
forts. Such efforts to control cue exposure, even though not sus-
tainable themselves, could be effective to the extent that they
provide people with opportunities to learn new responses.
An alternative to effortful, deliberate control of cue exposure is
provided by serendipitous changes in performance contexts that
occur naturally as a function of life events. As illustrated in Wood
et al.’s (2005) study of students transferring to a new university,
when the transfer involved change in the cues that triggered habits,
habit performance was disrupted, and students’ responses came
under the control of their behavioral goals instead of their habits.
By removing habit cues, the changes in performance circum-
stances promoted performance in line with people’s goals. Build-
ing on these findings, Verplanken and Wood (2006) proposed that
naturally occurring changes in performance contexts such as mov-
ing houses or changing jobs can be treated as opportunities for
habit change interventions. If people are best able to act on their
goals when related habits are disrupted, then it is during these
times that people’s overt responses are most likely to be vulnerable
to change through persuasive messages and other informational
interventions. The logic is to apply behavior change interventions
when people are best able to respond. Illustrating how this might
work, local governments seeking to increase use of public trans-
port would target new residents who have yet to establish car use
habits in their new locale and who are likely to be most susceptible
to the information provided.
In summary, interventions to break habits are best tailored to
address the processes of habitual responding to cues. In this view,
the often-seeming intractability of habitual behavior is partly a
product of interventions that fail to accommodate the close depen-
dence of habits on the contexts in which they are performed.
Habits are not easily changed through persuasive appeals that
target people’s goals. Instead, interventions to maximize habit
change provide people with concrete tools for controlling habit
cuing. One possibility is to exert short-term, effortful inhibition
that plausibly creates a window of opportunity for establishing
new, goal-consistent patterns of response. Another possibility is to
alter or avoid exposure to cues, a strategy that can involve delib-
erate decision making or exploiting serendipitous changes in per-
formance contexts. We see rich opportunities for development of
these approaches to the control of habit cuing in behavior change
In concluding this article, we note that despite promising initial
returns, our new habit model is still in its early days. In this final
section, we highlighted some new questions that flow from our
approach. These include the variety of ways that goals interface
with habit responding, in particular the possibility that goals might
co-opt habits that serve similar ends to the goals. We speculated
that this interface is made possible by the lack of a competing goal
representation in context–response associations that comprise hab-
its, which renders habits potentially compatible with multiple
goals. We also raised the possibility that people can regulate habit
performance through conflict-detection mechanisms involving the
ACC that do not require the activation of goals and the matching
of response outcomes to those goals. Although conflict signaled
through competing response tendencies provides a promising
mechanism for habit regulation, the specific application to habits
has yet to be developed. We also raised questions about the
conditions under which response repetition yields automatic goal
pursuit as opposed to habits. We speculated that a variety of factors
preserve the active role of goals in guiding performance during
repeated responding, including response complexity, equifinality,
and attention to action. Finally, we considered the implications of
our model for interventions to change habits. We proposed that
these are most effective when they address the cuing of habit
performance, either through inhibiting habit responses once acti-
vated or avoiding or altering exposure to the cues. The current
model provides a starting point for launching investigations into
these questions.
More generally, we have articulated a view of habits that moves
beyond the behaviorists’ conception of simple S-R associations
and places the habit construct within the broader structure of goal
pursuit. In this approach, the habit construct retains its rigid,
context-cued nature yet also interfaces with goals in ways that
allow for mutual influence and for habits to be regulated in line
with goals. In delineating the habit disposition in this way, our
model provides a framework for understanding, predicting, and
changing that unique component of everyday life in which behav-
ioral control has been outsourced directly onto the context cues
contiguous with past performance.
Aarts, H., & Dijksterhuis, A. (2000). Habits as knowledge structures:
Automaticity in goal-directed behavior. Journal of Personality and
Social Psychology, 78, 53– 63.
Aarts, H., & Dijksterhuis, A. (2003). The silence of the library: Environ-
ment, situational norm and social behavior. Journal of Personality and
Social Psychology, 84, 18 –28.
Adams, C. D. (1982). Variations in the sensitivity of instrumental respond-
ing to reinforcer devaluation. Quarterly Journal of Experimental Psy-
chology B, 34, 77–98.
Albarracı´n, D., Kumkale, G. T., & Johnson, B. T. (2002). Influences of
population and methodological factors on reasoning in condom use: A
meta-analysis. Unpublished manuscript, University of Florida.
Albarracı´n, D., McNatt, P. S., Klein, C. T. F., Ho, R. M., Mitchell, A. L.,
& Kumkale, G. T. (2003). Persuasive communications to change ac-
tions: An analysis of behavioral and cognitive impact in HIV prevention.
Health Psychology, 22, 166 –177.
Aldrich, J. A., Montgomery, J., & Wood, W. (2007, April). Repeated
turnout as a habit. Paper presented at the Midwestern Political Science
Society Meeting, Chicago.
Anderson, J. R. (1982). Acquisition of cognitive skill. Psychological Re-
view, 89, 369 406.
Baldwin, A. S., Rothman, A. J., Hertel, A. W., Linde, J. A., Jeffery, R. W.,
Finch, E. A., & Lando, H. A. (2006). Specifying the determinants of the
initiation and maintenance of behavior change: An examination of
self-efficacy, satisfaction, and smoking cessation. Health Psychology,
25, 626 634.
Banfield, J. F., Wyland, C. L., Macrae, C. N., Munte, T. F., & Heatherton,
T. F. (2004). The cognitive neuroscience of self-regulation. In R. F.
Baumeister & K. D. Vohs (Eds.), Handbook of self-regulation: Re-
search, theory, and applications (pp. 62–83). New York: Guilford Press.
Bargh, J. A. (1994). The Four Horsemen of automaticity: Awareness,
efficiency, intention, and control in social cognition. In R. S. Wyer Jr. &
T. K. Srull (Eds.), Handbook of social cognition (2nd ed., pp. 1–40).
Hillsdale, NJ: Erlbaum.
Bargh, J. A., & Barndollar, K. (1996). Automaticity in action: The uncon-
scious as repository of chronic goals and motives. In P. M. Gollwitzer &
J. A. Bargh (Eds.), The psychology of action: Linking cognition and
motivation to behavior (pp. 457– 481). New York: Guilford Press.
Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social
behavior: Direct effects of trait construct and stereotype activation on
action. Journal of Personality and Social Psychology, 71, 230 –244.
Bargh, J. A., & Ferguson, M. L. (2000). Beyond behaviorism: On the
automaticity of higher mental processes. Psychological Bulletin, 126,
Bargh, J. A., Gollwitzer, P. G., Lee-Chai, A., Barndollar, K., & Tro¨tschel,
R. (2001). The automated will: Nonconscious activation and pursuit of
behavioral goals. Journal of Personality and Social Psychology, 81,
1014 –1027.
Barker, R. G. (1968). Ecological psychology: Concepts and methods for
studying the environment of human behavior. Stanford, CA: Stanford
University Press.
Barker, R. G., & Schoggen, P. (1978). Measures of habitat and behavior
output. In R. G. Barker & Associates (Eds.), Habitats, environments, and
human behavior: Studies in ecological psychology and eco-behavioral
science from the Midwest Psychological Field Station, 1947–1972 (pp.
229 –244). San Francisco: Jossey-Bass.
Barnett, S. M., & Ceci, S. J. (2002). When and where do we apply what we
learn? A taxonomy for far transfer. Psychological Bulletin, 128, 612–
Baumeister, R. F. (1984). Choking under pressure: Self-consciousness and
paradoxical effects of incentives on skillful performance. Journal of
Personality and Social Psychology, 46, 610 620.
Baumeister, R. F., & Heatherton, T. D. (1996). Self-regulation failure: An
overview. Psychological Inquiry, 7, 1–15.
Bem, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances
in experimental social psychology (Vol. 6, pp. 1– 62). New York: Aca-
demic Press.
Betsch, T., Haberstroh, S., Molter, B., & Glo¨ckner, A. (2004). Oops, I did
it again—Relapse errors in routine decision making. Organizational
Behavior and Human Decision Processes, 93, 62–74.
Botvinick, M. M., Braaver, T. S., Barch, D. M., Carter, C. S., & Cohen,
J. D. (2001). Conflict monitoring and cognitive control. Psychological
Review, 108, 624 652.
Botvinick, M. M., & Bylsma, L. M. (2005). Distraction and action slips in
an everyday task: Evidence for a dynamic representation of task context.
Psychonomic Bulletin and Review, 12, 1011–1017.
Botvinick, M. M., & Plaut, D. C. (2004). Doing without schema hierar-
chies: A recurrent connectionist approach to normal and impaired rou-
tine sequential action. Psychological Review, 111, 395–429.
Botvinick, M. M., & Plaut, D. C. (2006). Such stuff as habits are made on:
A reply to Cooper and Shallice (2006). Psychological Review, 113,
Bouton, M. E. (2000). A learning theory perspective on lapse, relapse, and
the maintenance of behavior change. Health Psychology, 19(Suppl.),
57– 63.
Bower, G. H. (2000). A brief history of memory research. In E. Tulving &
F. I. M. Craik (Eds.), The Oxford handbook of memory (pp. 3–32). New
York: Oxford University Press.
Carver, C. S., & Scheier, M. F. (1998). On the self-regulation of behavior.
New York: Cambridge University Press.
Carver, C. S., & Scheier, M. F. (2002). Control processes and self-
organization as complementary principles underlying behavior. Person-
ality and Social Psychology Review, 6, 304–315.
Chaiken, S., & Trope. Y. (Eds.). (1999). Dual-process theories in social
psychology. New York: Guilford.
Chartrand, T. L., & Bargh, J. A. (1999). The chameleon effect: The
perception-behavior link and social interaction. Journal of Personality
and Social Psychology, 76, 893–910.
Chomsky, N. (1959). A review of B. F. Skinner’s “Verbal Behavior.”
Language, 35, 26 –58.
Cleeremans, A., Destrebecqz, A., & Boyer, M. (1998). Implicit learning:
News from the front. Trends in Cognitive Sciences, 2, 406 416.
Colwill, R. M., & Rescorla, R. A. (1988). The role of response-reinforcer
associations increases throughout extended instrumental training. Ani-
mal Learning & Behavior, 16, 105–111.
Cooper, R. P., & Shallice, T. (2006). Hierarchical schemas and goals in the
control of sequential behavior. Psychological Review, 113, 887–916.
Coutureau, E., & Killcross, S. (2003). Inactivation of the infralimbic
prefrontal cortex reinstates goal-directed responding in overtrained rats.
Behavioral Brain Research, 146, 167–174.
Custers, R., & Aarts, H. (2005). Positive affect as implicit motivator: On
the nonconscious operation of behavioral goals. Journal of Personality
and Social Psychology, 89, 129 –142.
Custers, R., & Aarts, H. (2007). In search of the nonconscious sources of
goal pursuit: Accessibility and positive affective valence of the goal
state. Journal of Experimental Social Psychology, 43, 312–318.
Daw, N. D., Niv, Y., & Dayan, P. (2005). Uncertainty based competition
between prefrontal and dorsolateral striatal systems for behavioral con-
trol. Nature Neuroscience, 8, 1704 –1711.
Dayan, P., & Balleine, B. W. (2002). Reward, motivation, and reinforce-
ment learning. Neuron, 36, 285–298.
Destrebecqz, A., & Cleeremans, A. (2001). Can sequence learning be
implicit? New evidence with the process dissociation procedure. Psy-
chonomic Bulletin and Review, 8, 343–350.
Dickinson, A., & Balleine, B. (2002). The role of learning in the operation
of motivational systems. In H. Pashler & R. Gallistel (Eds.), Stevens’
handbook of experimental psychology: Learning, motivation, and emo-
tion (3rd ed., Vol. 3, pp. 497–533). New York: Wiley.
Dickinson, A., Balleine, B., Watt, A., Gonzalez, F., & Boakes, R. A.
(1995). Motivational control after extended instrumental training. Ani-
mal Learning & Behavior, 23, 197–206.
Dijksterhuis, A., Bargh, J. A., & Miedema, J. (2000). Of men and mack-
erels: Attention and automatic behavior. In H. Bless & J. P. Forgas
(Eds.), Subjective experience in social cognition and behavior (pp.
36 –51). New York: Psychology Press.
Dijksterhuis, A., Spears, R., & Lepinasse, V. (2001). Reflecting and
deflecting stereotypes: Assimilation and contrast in impression forma-
tion and automatic behavior. Journal of Experimental Social Psychol-
ogy, 37, 286 –299.
Epley, N., Waytz, A., & Cacioppo, J. T. (2007). On seeing human: A
three-factor theory of anthropomorphism. Psychological Review, 114,
864 886.
Ferguson, E., & Bibby, P. A. (2002). Predicting future blood donor returns:
Past behavior, intentions, and observer effects. Health Psychology, 21,
Fishbach, A., & Trope, Y. (2005). The substitutability of external control
and self-control. Journal of Experimental Social Psychology, 41, 256
Floyd, D. L., Prentice-Dunn, S., & Rogers, R. W. (2000). A meta-analysis
of research on protection motivation theory. Journal of Applied Social
Psychology, 30, 407– 429.
Follette, W. C., & Hayes, S. C. (2001). Contemporary behavior therapy. In
C. R. Snyder & R. Ingram (Eds.), Handbook of psychological change:
Psychotherapy process and practices for the 21st century (pp. 381–408).
New York: Wiley.
Fo¨rster, J., Higgins, E. T., & Bianco, A. T. (2003). Speed/accuracy deci-
sions in task performance: Built-in trade-off or separate strategic con-
cerns? Organizational Behavior & Human Decision Processes, 90,
148 –164.
Freitas, A. L., Liberman, N., & Higgins, E. T. (2002). Regulatory fit and
resisting temptation during goal pursuit. Journal of Experimental Social
Psychology, 38, 291–298.
Frensch, P. A., & Ru¨nger, D. (2003). Implicit learning. Current Directions
in Psychological Science, 12, 13–18.
Gehring, W. J., Goss, B., Coles, M. G. H., Meyer, D. E., & Donchin, E.
(1993). A neural network system for error detection and compensation.
Psychological Science, 4, 385–390.
Glanz, K., Rimer, B. K., & Lewis, F. M. (2002). Health behavior and
health education: Theory, research and practice. San Francisco: Jossey-
Gollwitzer, P. M. (1999). Implementation intentions: Strong effects of
simple plans. American Psychologist, 54, 493–503.
Graybiel, A. (1998). The basal ganglia and chunking of action repertoires.
Neurobiology of Learning and Memory, 70, 119–136.
Gupta, P., & Cohen, N. J. (2002). Theoretical and computational analysis
of skill learning, repetition priming, and procedural memory. Psycho-
logical Review, 109, 401– 448.
Hay, J. F., & Jacoby, L. L. (1996). Separating habit and recollection:
Memory slips, process dissociations, and probability matching. Journal
of Experimental Psychology: Learning, Memory, and Cognition, 22,
Heatherton, T., & Nichols, P. (1994). Personal accounts of successful
versus failed attempts at life change. Personality and Social Psychology
Bulletin, 20, 664 675.
Hebb, D. O. (1949). The organization of behavior: A neuropsychological
theory. Oxford, England: Wiley.
Higgins, E. T. (1997). Beyond pleasure and pain. American Psychologist,
52, 1280 –1300.
Holland, R. W., Aarts, H., & Langendam, D. (2006). Breaking and creating
habits on the working floor: A field-experiment on the power of imple-
mentation intentions. Journal of Experimental Social Psychology, 42,
776 –783.
James, W. J. (1890). The principles of psychology. New York: Dover.
Ji, M., & Wood, W. (in press). Habitual purchase and consumption: Habits
and intentions guide behavior. Journal of Consumer Psychology.
Jime´nez, L., Me´ndez, C., & Cleeremans, A. (1996). Comparing direct and
indirect measures of sequence learning. Journal of Experimental Psy-
chology: Learning, Memory and Cognition, 22, 948–969.
Jonides, J. (2004). How does practice makes perfect? Nature Neuroscience,
7, 75–79.
Jueptner, M., Stephan, K. M., Frith, C. D., Brooks, D. J., Frackowiak,
R. S. J., & Passingham, R. E. (1997). Anatomy of motor learning. I.
Frontal cortex and attention to action. Journal of Neurophysiology, 77,
Kalivas, P. W., & McFarland, K. (2003). Brain circuitry and the reinstate-
ment of cocaine-seeking behavior. Psychopharmacology, 168, 44–56.
Kelly, A. M. C., & Garavan, H. (2005). Human functional neuroimaging of
brain changes associated with practice. Cerebral Cortex, 15, 1089
Klein, S. B., German, T. P., Cosmides, L., & Gabriel, R. (2004). A theory
of autobiographical memory: Necessary components and disorders re-
sulting from their loss. Social Cognition, 22, 460 490.
Klo¨ckner, C. A., Matthies, E., & Hunecke, M. (2003). Problems in opera-
tionalizing habits and integrating habits in normative decision-making
models. Journal of Applied Social Psychology, 33, 396 417.
Knowlton, B. J., Mangels, J. A., & Squire, L. R. (1996, September). A
neostriatal habit learning system in humans. Nature, 273, 1399 –1402.
Knutson, B., & Cooper, J. C. (2005). Functional magnetic resonance
imaging of reward prediction. Current Opinion in Neurology, 18, 411–
Kruglanski, A. W., Shah, J. Y., Fishbach, A., Friedman, R., Chun, W., &
Sleeth-Keppler, D. (2002). A theory of goal systems. In M. P. Zanna
(Ed.), Advances in experimental social psychology (Vol. 34, pp. 331–
378). New York: Academic Press.
LaBar, K. S., & Phelps, E. A. (2005). Reinstatement of conditioned fear in
humans is context-dependent and impaired in amnesia. Behavioral Neu-
roscience, 119, 677– 686.
Lewicki, P., Hill, T., & Bizot, E. (1988). Acquisition of procedural knowl-
edge about a pattern of stimuli that cannot be articulated. Cognitive
Psychology, 20, 24 –37.
Lieberman, M. D., & Eisenberger, N. I. (2004). Conflict and habit: A social
cognitive neuroscience approach to the self. In A. Tesser, J. V. Wood, &
D. A. Stapel (Eds.), On building, defending and regulating the self: A
psychological perspective (pp. 77–102). New York: Psychology Press.
Lieberman, M. D., Jarcho, J. M., & Satpute, A. B. (2004). Evidence-based
and intuition-based self-knowledge: An fMRI study. Journal of Person-
ality and Social Psychology, 87, 421–435.
Loewenstein, G. (1996). Out of control: Visceral influences on behavior.
Organizational Behavior and Human Decision Processes, 65, 272–292.
Macrae, C. N., & Johnston, L. (1998). Help, I need somebody: Automatic
action and inaction. Social Cognition, 16, 400 417.
Mandler, G. (2002). Organisation: What levels of processing are levels of.
Memory, 10, 333–338.
Mayr, U., & Bryck, R. L. (2005). Sticky rules: Integration between abstract
rules and specific actions. Journal of Experimental Psychology: Learn-
ing, Memory, and Cognition, 31, 337–350.
McClelland, J. L., McNaughton, B. L., & O’Reilly, R. C. (1995). Why
there are complementary learning systems in the hippocampus and
neocortex: Insights from the successes and failures of connectionist
models of learning and memory. Psychological Review, 102, 419 457.
Metcalfe, J., & Mischel, W. (1999). A hot/cool-system analysis of delay of
gratification: Dynamics of willpower. Psychological Review, 106, 3–19.
Miller, E. K., & Cohen, J. D. (2001). An integrative theory of prefrontal
cortex function. Annual Review of Neuroscience, 24, 167–202.
Miller, G. A., Galanter, E., & Pribram, K. H. (1960). Plans and the
structure of behavior. New York: Henry Holt.
Mirenowicz, J., & Schultz, W. (1996, February 1). Preferential activation
of midbrain dopamine neurons by appetitive rather than aversive stimuli.
Nature, 379, 449 451.
Montague, P. R., Hyman, S. E., & Cohen, J. D. (2004, October 14).
Computational roles for dopamine in behavioural control. Nature, 431,
760 –767.
Moors, A., & De Houwer, J. (2006). Automaticity: A theoretical and
conceptual analysis. Psychological Bulletin, 132, 297–326.
Mowrer, O. H. (1960). Learning theory and the symbolic processes. New
York: Wiley.
Muraven, M., & Baumeister, R. F. (2000). Self-regulation and depletion of
limited resources: Does self-control resemble a muscle? Psychological
Bulletin, 126, 247–259.
Neal, D. T., Pascoe, A., & Wood, W. (2007). Effortless perfection: Par-
adoxical effects of explicit and implicit goals on habitual responding.
Unpublished manuscript, Duke University, Durham, NC.
Neal, D. T., & Wood, W. (2007). Automaticity in situ: Direct context cuing
of habits in daily life. In J. A. Bargh, P. Gollwitzer, & E. Morsella (Eds.),
Psychology of action (Vol. 2): Mechanisms of human action. London:
Oxford University Press.
Neal, D. T., Wood, W., & Quinn, J. M. (2006). Habits: A repeat perfor-
mance. Current Directions in Psychological Science, 15, 198 –202.
Neisser, U. (1967). Cognitive psychology. New York: Appleton Century-
Nisbett, R., & Wilson, T. D. (1977). Telling more than we can know:
Verbal reports on mental processes. Psychological Review, 84, 231–259.
Norman, D. A. (1981). Categorization of action slips. Psychological Re-
view, 88, 1–15.
Norman, D. A., & Shallice, T. (1986). Attention to action: Willed and
automatic control of behavior. In R. J. Davidson, G. E. Schwartz, & D.
Shapiro (Eds.), Consciousness and self-regulation (Vol. 4, pp. 1–18).
New York: Plenum.
O’Doherty, J. P. (2004). Reward representations and reward-related learn-
ing in the human brain: Insights from neuroimaging. Current Opinion in
Neurobiology, 14, 769 –776.
Ouellette, J. A., & Wood, W. (1998). Habit and intention in everyday life:
The multiple processes by which past behavior predicts future behavior.
Psychological Bulletin, 124, 54 –74.
Packard, M. G. (1999). Glutamate infused post-training into the hippocam-
pus or caudate-putamen differentially strengthens place and response
learning. Proceedings of the National Academy of Sciences, 96, 12881–
Packard, M. G., & Knowlton, B. J. (2002). Learning and memory functions
of the basal ganglia. Annual Review of Neuroscience, 25, 563–593.
Packard, M. G., & McGaugh, J. L. (1996). Inactivation of hippocampus or
caudate nucleus with lidocaine differentially affects expression of place
and response learning. Neurobiology of Learning and Memory, 65,
Pasupathy, A., & Miller, E. K. (2005). Different time courses for learning-
related activity in the prefrontal cortex and striatum. Nature, 433, 873–
Pessiglione, M., Schmidt, L., Draganski, B., Kalisch, R., Lau, H., Dolan,
R. J., & Frith, C. D. (2007, May 11). How the brain translates money
into force: A neuroimaging study of subliminal motivation. Science, 316,
904 –906.
Poldrack, R. A., Clark, J., Pa´re-Blagoev, J., Shohamy, D., Creso Moyano,
J., Myers, C., & Gluck, M. A. (2001, November). Interactive memory
systems in the human brain. Nature, 414, 546 –550.
Polivy, J. (1998). The effects of behavioral inhibition: Integrating internal
cues, cognition, behavior, and affect. Psychological Inquiry, 9, 181–204.
Prinz, W. (1990). A common coding approach to perception and action. In
O. Neumann & W. Prinz (Eds.), Relationships between perception and
action: Current approaches (pp. 167–201). Berlin: Springer-Verlag.
Prochaska, J. O., DiClemente, C. C., & Norcross, J. C. (1992). In search of
how people change: Applications to addictive behaviors. American
Psychologist, 47, 1102–1114.
Quinn, J. M., Pascoe, A., & Wood, W. (2007). Self-control of habits and
temptations: Data from the front lines of behavior change. Manuscript
under review.
Quinn, J. M., & Wood, W. (2005). Habits across the lifespan. Unpublished
manuscript, Duke University, Durham, NC.
Reason, J. T. (1990). Human error. Cambridge, England: Cambridge
University Press.
Reason, J. T. (1992). Cognitive underspecification: Its varieties and con-
sequences. In B. J. Baars (Ed.), Experimental slips and human error:
Exploring the architecture of volition (pp. 71–91). New York: Plenum
Rizzolatti, G., Fadiga, L., Gallese, V., & Fogassi, L. (1996). Premotor
cortex and the recognition of motor actions. Cognitive Brain Research,
3, 131–141.
Rowe, J., Friston, K., Frackowiak, R., & Passingham, R. (2002). Attention
to action: Specific modulation of corticocortical interactions in humans.
Neuroimage, 17, 988 –998.
Schmajuk, N. A., Larrauri, J. A., & LaBar, K. S. (2007). Reinstatement of
conditioned fear and the hippocampus: An attentional-associative
model. Behavioral and Brain Research, 177, 242–253.
Schultz, W. (2006). Behavioral theories and the neurophysiology of re-
ward. Annual Review of Psychology, 57, 87–115.
Shah, J. Y., & Kruglanski, A. W. (2002). Priming against your will: How
accessible alternatives affect goal pursuit. Journal of Experimental So-
cial Psychology, 38, 368 –383.
Shanks, D. R., & St. John, M. F. (1994). Characteristics of dissociable
human learning systems. Behavioral and Brain Sciences, 17, 367– 447.
Sheeran, P., Aarts, H., Custers, R., Rivis, A., Webb, T. L., & Cooke, R.
(2005). The goal-dependent automaticity of drinking habits. British
Journal of Social Psychology, 44, 47–63.
Sheeran, P., Webb, T. L., & Gollwitzer, P. (2005). The interplay between
goal intentions and implementation intentions. Personality and Social
Psychology Bulletin, 31, 87–98.
Simonson, I. (2005). In defense of consciousness: The role of conscious
and unconscious inputs in consumer choice. Journal of Consumer Psy-
chology, 15, 211–217.
Skinner, B. F. (1938). The behavior of organisms: An experimental anal-
ysis. New York: Appleton-Century.
Smith, E. R., & DeCoster, J. (2000). Dual-process models in social and
cognitive psychology: Conceptual integration and links to underlying
memory systems. Personality and Social Psychology Review, 4, 108
Sobal, J., & Wansink, B. (2007). Kitchenscapes, tablescapes, platescapes,
and foodscapes: Influence of microscale built environments on food
intake. Environment and Behavior, 39, 124–142.
Sun, R., Slusarz, P., & Terry, C. (2005). The interaction of the explicit and
the implicit in skill learning: A dual-process approach. Psychological
Review, 112, 159 –192.
Thorndike, E. L. (1898). Animal intelligence: An experimental study of the
associative processes in animals. Psychological Review Monograph
Supplement, 2, 1–109.
Tice, D. M., Butler, J. L., Muraven, M. B., & Stillwell, A. M. (1995). When
modesty prevails: Differential favorability of self-presentation to friends
and strangers. Journal of Personality and Social Psychology, 69, 1120
Townsend, D. J., & Bever, T. G. (2001). Sentence comprehension: The
integration of habits and rules. Cambridge, MA: MIT Press.
Tucker, M., & Ellis, R. (2004). Action priming by briefly presented
objects. Acta Psychologica, 116, 185–203.
Verplanken, B., & Aarts, H. (1999). Habit, attitude, and planned beh